id	sid	tid	token	lemma	pos
ijassa-372	1	1	advances	advance	NOUN
ijassa-372	1	2	in	in	ADP
ijassa-372	1	3	systems	system	NOUN
ijassa-372	1	4	science	science	NOUN
ijassa-372	1	5	and	and	CCONJ
ijassa-372	1	6	application	application	NOUN
ijassa-372	1	7	(	(	PUNCT
ijassa-372	1	8	2015	2015	NUM
ijassa-372	1	9	)	)	PUNCT
ijassa-372	1	10	vol.15	vol.15	NOUN
ijassa-372	1	11	no.3	no.3	VERB
ijassa-372	1	12	242	242	NUM
ijassa-372	1	13	-	-	SYM
ijassa-372	1	14	253	253	NUM
ijassa-372	1	15	image	image	NOUN
ijassa-372	1	16	retrieval	retrieval	NOUN
ijassa-372	1	17	based	base	VERB
ijassa-372	1	18	on	on	ADP
ijassa-372	1	19	spatial	spatial	ADJ
ijassa-372	1	20	organization	organization	NOUN
ijassa-372	1	21	of	of	ADP
ijassa-372	1	22	patches	patches	PROPN
ijassa-372	1	23	yan	yan	PROPN
ijassa-372	1	24	yu1,2	yu1,2	PROPN
ijassa-372	1	25	1hubei	1hubei	NUM
ijassa-372	1	26	province	province	NOUN
ijassa-372	1	27	key	key	ADJ
ijassa-372	1	28	laboratory	laboratory	NOUN
ijassa-372	1	29	of	of	ADP
ijassa-372	1	30	systems	system	NOUN
ijassa-372	1	31	science	science	NOUN
ijassa-372	1	32	in	in	ADP
ijassa-372	1	33	metallurgical	metallurgical	ADJ
ijassa-372	1	34	process	process	NOUN
ijassa-372	1	35	,	,	PUNCT
ijassa-372	1	36	wuhan	wuhan	PROPN
ijassa-372	1	37	university	university	PROPN
ijassa-372	1	38	of	of	ADP
ijassa-372	1	39	science	science	NOUN
ijassa-372	1	40	and	and	CCONJ
ijassa-372	1	41	technology	technology	NOUN
ijassa-372	1	42	,	,	PUNCT
ijassa-372	1	43	china	china	PROPN
ijassa-372	1	44	2school	2school	NUM
ijassa-372	1	45	of	of	ADP
ijassa-372	1	46	science	science	NOUN
ijassa-372	1	47	,	,	PUNCT
ijassa-372	1	48	wuhan	wuhan	PROPN
ijassa-372	1	49	university	university	PROPN
ijassa-372	1	50	of	of	ADP
ijassa-372	1	51	science	science	NOUN
ijassa-372	1	52	and	and	CCONJ
ijassa-372	1	53	technology	technology	NOUN
ijassa-372	1	54	,	,	PUNCT
ijassa-372	1	55	china	china	PROPN
ijassa-372	1	56	abstract	abstract	ADJ
ijassa-372	1	57	this	this	DET
ijassa-372	1	58	paper	paper	NOUN
ijassa-372	1	59	presents	present	VERB
ijassa-372	1	60	a	a	DET
ijassa-372	1	61	novel	novel	ADJ
ijassa-372	1	62	approach	approach	NOUN
ijassa-372	1	63	for	for	ADP
ijassa-372	1	64	image	image	NOUN
ijassa-372	1	65	retrieval	retrieval	NOUN
ijassa-372	1	66	by	by	ADP
ijassa-372	1	67	utilizing	utilize	VERB
ijassa-372	1	68	the	the	DET
ijassa-372	1	69	spatial	spatial	ADJ
ijassa-372	1	70	organization	organization	NOUN
ijassa-372	1	71	of	of	ADP
ijassa-372	1	72	image	image	NOUN
ijassa-372	1	73	patches	patch	NOUN
ijassa-372	1	74	.	.	PUNCT
ijassa-372	2	1	firstly	firstly	ADV
ijassa-372	2	2	,	,	PUNCT
ijassa-372	2	3	each	each	DET
ijassa-372	2	4	image	image	NOUN
ijassa-372	2	5	is	be	AUX
ijassa-372	2	6	presented	present	VERB
ijassa-372	2	7	as	as	ADP
ijassa-372	2	8	a	a	DET
ijassa-372	2	9	set	set	NOUN
ijassa-372	2	10	of	of	ADP
ijassa-372	2	11	patches	patch	NOUN
ijassa-372	2	12	and	and	CCONJ
ijassa-372	2	13	each	each	DET
ijassa-372	2	14	patch	patch	NOUN
ijassa-372	2	15	is	be	AUX
ijassa-372	2	16	descripted	descripte	VERB
ijassa-372	2	17	as	as	ADP
ijassa-372	2	18	a	a	DET
ijassa-372	2	19	bag	bag	NOUN
ijassa-372	2	20	-	-	PUNCT
ijassa-372	2	21	of	of	ADP
ijassa-372	2	22	-	-	PUNCT
ijassa-372	2	23	visual	visual	ADJ
ijassa-372	2	24	-	-	PUNCT
ijassa-372	2	25	words	word	NOUN
ijassa-372	2	26	(	(	PUNCT
ijassa-372	2	27	bow	bow	NOUN
ijassa-372	2	28	)	)	PUNCT
ijassa-372	2	29	feature	feature	NOUN
ijassa-372	2	30	.	.	PUNCT
ijassa-372	3	1	secondly	secondly	ADV
ijassa-372	3	2	,	,	PUNCT
ijassa-372	3	3	the	the	DET
ijassa-372	3	4	matrix	matrix	NOUN
ijassa-372	3	5	of	of	ADP
ijassa-372	3	6	matching	matching	NOUN
ijassa-372	3	7	costs	cost	NOUN
ijassa-372	3	8	between	between	ADP
ijassa-372	3	9	all	all	DET
ijassa-372	3	10	pairs	pair	NOUN
ijassa-372	3	11	of	of	ADP
ijassa-372	3	12	patches	patch	NOUN
ijassa-372	3	13	that	that	PRON
ijassa-372	3	14	come	come	VERB
ijassa-372	3	15	from	from	ADP
ijassa-372	3	16	the	the	DET
ijassa-372	3	17	two	two	NUM
ijassa-372	3	18	images	image	NOUN
ijassa-372	3	19	respectively	respectively	ADV
ijassa-372	3	20	is	be	AUX
ijassa-372	3	21	computed	compute	VERB
ijassa-372	3	22	by	by	ADP
ijassa-372	3	23	utilizing	utilize	VERB
ijassa-372	3	24	both	both	DET
ijassa-372	3	25	the	the	DET
ijassa-372	3	26	appearance	appearance	NOUN
ijassa-372	3	27	feature	feature	NOUN
ijassa-372	3	28	and	and	CCONJ
ijassa-372	3	29	spatial	spatial	ADJ
ijassa-372	3	30	information	information	NOUN
ijassa-372	3	31	.	.	PUNCT
ijassa-372	4	1	finally	finally	ADV
ijassa-372	4	2	,	,	PUNCT
ijassa-372	4	3	the	the	DET
ijassa-372	4	4	distance	distance	NOUN
ijassa-372	4	5	between	between	ADP
ijassa-372	4	6	two	two	NUM
ijassa-372	4	7	images	image	NOUN
ijassa-372	4	8	is	be	AUX
ijassa-372	4	9	obtained	obtain	VERB
ijassa-372	4	10	by	by	ADP
ijassa-372	4	11	minimize	minimize	VERB
ijassa-372	4	12	the	the	DET
ijassa-372	4	13	total	total	ADJ
ijassa-372	4	14	cost	cost	NOUN
ijassa-372	4	15	of	of	ADP
ijassa-372	4	16	matching	match	VERB
ijassa-372	4	17	using	use	VERB
ijassa-372	4	18	hungarian	hungarian	ADJ
ijassa-372	4	19	algorithm	algorithm	NOUN
ijassa-372	4	20	.	.	PUNCT
ijassa-372	5	1	the	the	DET
ijassa-372	5	2	resulting	result	VERB
ijassa-372	5	3	distance	distance	NOUN
ijassa-372	5	4	is	be	AUX
ijassa-372	5	5	applied	apply	VERB
ijassa-372	5	6	to	to	ADP
ijassa-372	5	7	image	image	NOUN
ijassa-372	5	8	retrieval	retrieval	NOUN
ijassa-372	5	9	.	.	PUNCT
ijassa-372	6	1	the	the	DET
ijassa-372	6	2	experimental	experimental	ADJ
ijassa-372	6	3	results	result	NOUN
ijassa-372	6	4	indicate	indicate	VERB
ijassa-372	6	5	that	that	SCONJ
ijassa-372	6	6	the	the	DET
ijassa-372	6	7	proposed	propose	VERB
ijassa-372	6	8	method	method	NOUN
ijassa-372	6	9	outperforms	outperform	VERB
ijassa-372	6	10	the	the	DET
ijassa-372	6	11	traditional	traditional	ADJ
ijassa-372	6	12	bow	bow	NOUN
ijassa-372	6	13	method	method	NOUN
ijassa-372	6	14	in	in	ADP
ijassa-372	6	15	image	image	NOUN
ijassa-372	6	16	retrieval	retrieval	NOUN
ijassa-372	6	17	scenario	scenario	NOUN
ijassa-372	6	18	.	.	PUNCT
ijassa-372	7	1	keywords	keyword	VERB
ijassa-372	7	2	image	image	NOUN
ijassa-372	7	3	retrieval	retrieval	NOUN
ijassa-372	7	4	;	;	PUNCT
ijassa-372	7	5	bag	bag	NOUN
ijassa-372	7	6	-	-	PUNCT
ijassa-372	7	7	of	of	ADP
ijassa-372	7	8	-	-	PUNCT
ijassa-372	7	9	visual	visual	ADJ
ijassa-372	7	10	-	-	PUNCT
ijassa-372	7	11	words	word	NOUN
ijassa-372	7	12	;	;	PUNCT
ijassa-372	7	13	image	image	NOUN
ijassa-372	7	14	patch	patch	NOUN
ijassa-372	7	15	;	;	PUNCT
ijassa-372	7	16	image	image	NOUN
ijassa-372	7	17	distance	distance	NOUN
ijassa-372	7	18	measure	measure	NOUN
ijassa-372	7	19	;	;	PUNCT
ijassa-372	7	20	hungarian	hungarian	ADJ
ijassa-372	7	21	algorithm	algorithm	NOUN
ijassa-372	7	22	.	.	PUNCT
ijassa-372	8	1	1	1	NUM
ijassa-372	8	2	introduction	introduction	NOUN
ijassa-372	8	3	the	the	DET
ijassa-372	8	4	most	most	ADV
ijassa-372	8	5	popular	popular	ADJ
ijassa-372	8	6	approaches	approach	NOUN
ijassa-372	8	7	for	for	ADP
ijassa-372	8	8	content	content	NOUN
ijassa-372	8	9	-	-	PUNCT
ijassa-372	8	10	based	base	VERB
ijassa-372	8	11	image	image	NOUN
ijassa-372	8	12	retrieval	retrieval	NOUN
ijassa-372	8	13	nowadays	nowadays	ADV
ijassa-372	8	14	are	be	AUX
ijassa-372	8	15	based	base	VERB
ijassa-372	8	16	on	on	ADP
ijassa-372	8	17	the	the	DET
ijassa-372	8	18	model	model	NOUN
ijassa-372	8	19	of	of	ADP
ijassa-372	8	20	bag	bag	NOUN
ijassa-372	8	21	-	-	PUNCT
ijassa-372	8	22	of	of	ADP
ijassa-372	8	23	-	-	PUNCT
ijassa-372	8	24	visual	visual	ADJ
ijassa-372	8	25	-	-	PUNCT
ijassa-372	8	26	words	word	NOUN
ijassa-372	8	27	(	(	PUNCT
ijassa-372	8	28	bow	bow	NOUN
ijassa-372	8	29	)	)	PUNCT
ijassa-372	8	30	.	.	PUNCT
ijassa-372	9	1	they	they	PRON
ijassa-372	9	2	represent	represent	VERB
ijassa-372	9	3	image	image	NOUN
ijassa-372	9	4	as	as	ADP
ijassa-372	9	5	an	an	DET
ijassa-372	9	6	orderless	orderless	NOUN
ijassa-372	9	7	collection	collection	NOUN
ijassa-372	9	8	of	of	ADP
ijassa-372	9	9	local	local	ADJ
ijassa-372	9	10	features	feature	NOUN
ijassa-372	9	11	and	and	CCONJ
ijassa-372	9	12	demonstrate	demonstrate	VERB
ijassa-372	9	13	impressive	impressive	ADJ
ijassa-372	9	14	performance	performance	NOUN
ijassa-372	9	15	because	because	SCONJ
ijassa-372	9	16	of	of	ADP
ijassa-372	9	17	their	their	PRON
ijassa-372	9	18	compactness	compactness	NOUN
ijassa-372	9	19	and	and	CCONJ
ijassa-372	9	20	invariance	invariance	NOUN
ijassa-372	9	21	.	.	PUNCT
ijassa-372	10	1	the	the	DET
ijassa-372	10	2	traditional	traditional	ADJ
ijassa-372	10	3	approach	approach	NOUN
ijassa-372	10	4	based	base	VERB
ijassa-372	10	5	on	on	ADP
ijassa-372	10	6	bow	bow	NOUN
ijassa-372	10	7	contains	contain	VERB
ijassa-372	10	8	the	the	DET
ijassa-372	10	9	following	follow	VERB
ijassa-372	10	10	steps	step	NOUN
ijassa-372	10	11	.	.	PUNCT
ijassa-372	11	1	firstly	firstly	ADV
ijassa-372	11	2	,	,	PUNCT
ijassa-372	11	3	local	local	ADJ
ijassa-372	11	4	features	feature	NOUN
ijassa-372	11	5	are	be	AUX
ijassa-372	11	6	extracted	extract	VERB
ijassa-372	11	7	from	from	ADP
ijassa-372	11	8	a	a	DET
ijassa-372	11	9	training	training	NOUN
ijassa-372	11	10	set	set	NOUN
ijassa-372	11	11	of	of	ADP
ijassa-372	11	12	images	image	NOUN
ijassa-372	11	13	by	by	ADP
ijassa-372	11	14	interesting	interesting	ADJ
ijassa-372	11	15	point	point	NOUN
ijassa-372	11	16	sampling	sample	VERB
ijassa-372	11	17	or	or	CCONJ
ijassa-372	11	18	dense	dense	ADJ
ijassa-372	11	19	sampling	sampling	NOUN
ijassa-372	11	20	.	.	PUNCT
ijassa-372	12	1	secondly	secondly	ADV
ijassa-372	12	2	,	,	PUNCT
ijassa-372	12	3	all	all	DET
ijassa-372	12	4	the	the	DET
ijassa-372	12	5	feature	feature	NOUN
ijassa-372	12	6	vectors	vector	NOUN
ijassa-372	12	7	are	be	AUX
ijassa-372	12	8	quantized	quantize	VERB
ijassa-372	12	9	to	to	PART
ijassa-372	12	10	construct	construct	VERB
ijassa-372	12	11	a	a	DET
ijassa-372	12	12	dictionary	dictionary	NOUN
ijassa-372	12	13	of	of	ADP
ijassa-372	12	14	visual	visual	ADJ
ijassa-372	12	15	words	word	NOUN
ijassa-372	12	16	by	by	ADP
ijassa-372	12	17	k	k	ADJ
ijassa-372	12	18	-	-	PUNCT
ijassa-372	12	19	means	means	NOUN
ijassa-372	12	20	clustering	clustering	NOUN
ijassa-372	12	21	.	.	PUNCT
ijassa-372	13	1	each	each	DET
ijassa-372	13	2	cluster	cluster	NOUN
ijassa-372	13	3	tends	tend	VERB
ijassa-372	13	4	to	to	PART
ijassa-372	13	5	contain	contain	VERB
ijassa-372	13	6	image	image	NOUN
ijassa-372	13	7	patches	patch	NOUN
ijassa-372	13	8	with	with	ADP
ijassa-372	13	9	similar	similar	ADJ
ijassa-372	13	10	appearance	appearance	NOUN
ijassa-372	13	11	and	and	CCONJ
ijassa-372	13	12	its	its	PRON
ijassa-372	13	13	centroid	centroid	NOUN
ijassa-372	13	14	represents	represent	VERB
ijassa-372	13	15	a	a	DET
ijassa-372	13	16	visual	visual	ADJ
ijassa-372	13	17	word	word	NOUN
ijassa-372	13	18	.	.	PUNCT
ijassa-372	14	1	thirdly	thirdly	ADV
ijassa-372	14	2	,	,	PUNCT
ijassa-372	14	3	each	each	DET
ijassa-372	14	4	image	image	NOUN
ijassa-372	14	5	is	be	AUX
ijassa-372	14	6	descripted	descripte	VERB
ijassa-372	14	7	as	as	ADP
ijassa-372	14	8	a	a	DET
ijassa-372	14	9	histogram	histogram	NOUN
ijassa-372	14	10	indicating	indicate	VERB
ijassa-372	14	11	the	the	DET
ijassa-372	14	12	probability	probability	NOUN
ijassa-372	14	13	distribution	distribution	NOUN
ijassa-372	14	14	of	of	ADP
ijassa-372	14	15	visual	visual	ADJ
ijassa-372	14	16	words	word	NOUN
ijassa-372	14	17	by	by	ADP
ijassa-372	14	18	hard	hard	ADJ
ijassa-372	14	19	assignment	assignment	NOUN
ijassa-372	14	20	or	or	CCONJ
ijassa-372	14	21	soft	soft	ADJ
ijassa-372	14	22	assignment	assignment	NOUN
ijassa-372	14	23	.	.	PUNCT
ijassa-372	15	1	lastly	lastly	ADV
ijassa-372	15	2	,	,	PUNCT
ijassa-372	15	3	the	the	DET
ijassa-372	15	4	distances	distance	NOUN
ijassa-372	15	5	between	between	ADP
ijassa-372	15	6	histograms	histogram	NOUN
ijassa-372	15	7	are	be	AUX
ijassa-372	15	8	used	use	VERB
ijassa-372	15	9	for	for	ADP
ijassa-372	15	10	image	image	NOUN
ijassa-372	15	11	retrieval	retrieval	NOUN
ijassa-372	15	12	.	.	PUNCT
ijassa-372	16	1	the	the	DET
ijassa-372	16	2	major	major	ADJ
ijassa-372	16	3	limitation	limitation	NOUN
ijassa-372	16	4	of	of	ADP
ijassa-372	16	5	such	such	ADJ
ijassa-372	16	6	approaches	approach	NOUN
ijassa-372	16	7	is	be	AUX
ijassa-372	16	8	that	that	SCONJ
ijassa-372	16	9	their	their	PRON
ijassa-372	16	10	image	image	NOUN
ijassa-372	16	11	representation	representation	NOUN
ijassa-372	16	12	ignores	ignore	VERB
ijassa-372	16	13	encoding	encode	VERB
ijassa-372	16	14	the	the	DET
ijassa-372	16	15	spatial	spatial	ADJ
ijassa-372	16	16	information	information	NOUN
ijassa-372	16	17	or	or	CCONJ
ijassa-372	16	18	geometrical	geometrical	ADJ
ijassa-372	16	19	relationships	relationship	NOUN
ijassa-372	16	20	among	among	ADP
ijassa-372	16	21	visual	visual	ADJ
ijassa-372	16	22	words	word	NOUN
ijassa-372	16	23	,	,	PUNCT
ijassa-372	16	24	which	which	PRON
ijassa-372	16	25	has	have	AUX
ijassa-372	16	26	been	be	AUX
ijassa-372	16	27	discovered	discover	VERB
ijassa-372	16	28	very	very	ADV
ijassa-372	16	29	important	important	ADJ
ijassa-372	16	30	for	for	ADP
ijassa-372	16	31	understanding	understand	VERB
ijassa-372	16	32	image	image	NOUN
ijassa-372	16	33	content	content	NOUN
ijassa-372	16	34	[	[	X
ijassa-372	16	35	1	1	NUM
ijassa-372	16	36	]	]	PUNCT
ijassa-372	16	37	.	.	PUNCT
ijassa-372	17	1	for	for	ADP
ijassa-372	17	2	solving	solve	VERB
ijassa-372	17	3	the	the	DET
ijassa-372	17	4	problem	problem	NOUN
ijassa-372	17	5	,	,	PUNCT
ijassa-372	17	6	some	some	DET
ijassa-372	17	7	existing	exist	VERB
ijassa-372	17	8	approaches	approach	NOUN
ijassa-372	17	9	leave	leave	VERB
ijassa-372	17	10	the	the	DET
ijassa-372	17	11	spatial	spatial	ADJ
ijassa-372	17	12	verification	verification	NOUN
ijassa-372	17	13	as	as	ADP
ijassa-372	17	14	a	a	DET
ijassa-372	17	15	post	post	ADJ
ijassa-372	17	16	-	-	ADJ
ijassa-372	17	17	processing	processing	ADJ
ijassa-372	17	18	step	step	NOUN
ijassa-372	17	19	[	[	X
ijassa-372	17	20	2	2	NUM
ijassa-372	17	21	-	-	SYM
ijassa-372	17	22	3	3	NUM
ijassa-372	17	23	]	]	PUNCT
ijassa-372	17	24	,	,	PUNCT
ijassa-372	17	25	however	however	ADV
ijassa-372	17	26	,	,	PUNCT
ijassa-372	17	27	they	they	PRON
ijassa-372	17	28	are	be	AUX
ijassa-372	17	29	only	only	ADV
ijassa-372	17	30	suitable	suitable	ADJ
ijassa-372	17	31	for	for	ADP
ijassa-372	17	32	partial	partial	ADJ
ijassa-372	17	33	-	-	PUNCT
ijassa-372	17	34	duplicate	duplicate	ADJ
ijassa-372	17	35	image	image	NOUN
ijassa-372	17	36	retrieval	retrieval	NOUN
ijassa-372	17	37	,	,	PUNCT
ijassa-372	17	38	as	as	ADP
ijassa-372	17	39	for	for	ADP
ijassa-372	17	40	semantic	semantic	ADJ
ijassa-372	17	41	retrieval	retrieval	NOUN
ijassa-372	17	42	they	they	PRON
ijassa-372	17	43	are	be	AUX
ijassa-372	17	44	limited	limited	ADJ
ijassa-372	17	45	.	.	PUNCT
ijassa-372	18	1	spatial	spatial	ADJ
ijassa-372	18	2	pyramid	pyramid	NOUN
ijassa-372	18	3	matching	match	VERB
ijassa-372	18	4	[	[	X
ijassa-372	18	5	4	4	NUM
ijassa-372	18	6	]	]	PUNCT
ijassa-372	18	7	is	be	AUX
ijassa-372	18	8	successful	successful	ADJ
ijassa-372	18	9	for	for	ADP
ijassa-372	18	10	representing	represent	VERB
ijassa-372	18	11	the	the	DET
ijassa-372	18	12	spatial	spatial	ADJ
ijassa-372	18	13	information	information	NOUN
ijassa-372	18	14	advances	advance	VERB
ijassa-372	18	15	in	in	ADP
ijassa-372	18	16	systems	system	NOUN
ijassa-372	18	17	science	science	NOUN
ijassa-372	18	18	and	and	CCONJ
ijassa-372	18	19	application	application	NOUN
ijassa-372	18	20	(	(	PUNCT
ijassa-372	18	21	2015	2015	NUM
ijassa-372	18	22	)	)	PUNCT
ijassa-372	18	23	vol.15	vol.15	NOUN
ijassa-372	18	24	no.3	no.3	VERB
ijassa-372	18	25	243	243	NUM
ijassa-372	18	26	of	of	ADP
ijassa-372	18	27	visual	visual	ADJ
ijassa-372	18	28	words	word	NOUN
ijassa-372	18	29	,	,	PUNCT
ijassa-372	18	30	which	which	PRON
ijassa-372	18	31	subsquently	subsquently	ADV
ijassa-372	18	32	turn	turn	VERB
ijassa-372	18	33	into	into	ADP
ijassa-372	18	34	a	a	DET
ijassa-372	18	35	general	general	ADJ
ijassa-372	18	36	framework	framework	NOUN
ijassa-372	18	37	combined	combine	VERB
ijassa-372	18	38	with	with	ADP
ijassa-372	18	39	various	various	ADJ
ijassa-372	18	40	feature	feature	NOUN
ijassa-372	18	41	encoding	encode	VERB
ijassa-372	18	42	methods	method	NOUN
ijassa-372	18	43	[	[	X
ijassa-372	18	44	5	5	NUM
ijassa-372	18	45	-	-	SYM
ijassa-372	18	46	6	6	NUM
ijassa-372	18	47	]	]	PUNCT
ijassa-372	18	48	for	for	ADP
ijassa-372	18	49	image	image	NOUN
ijassa-372	18	50	classification	classification	NOUN
ijassa-372	18	51	.	.	PUNCT
ijassa-372	19	1	for	for	ADP
ijassa-372	19	2	analogous	analogous	ADJ
ijassa-372	19	3	approaches	approach	NOUN
ijassa-372	19	4	like	like	ADP
ijassa-372	19	5	spm	spm	PROPN
ijassa-372	19	6	,	,	PUNCT
ijassa-372	19	7	we	we	PRON
ijassa-372	19	8	uniformly	uniformly	ADV
ijassa-372	19	9	call	call	VERB
ijassa-372	19	10	them	they	PRON
ijassa-372	19	11	patch	patch	NOUN
ijassa-372	19	12	-	-	PUNCT
ijassa-372	19	13	based	base	VERB
ijassa-372	19	14	methods	method	NOUN
ijassa-372	19	15	.	.	PUNCT
ijassa-372	20	1	in	in	ADP
ijassa-372	20	2	methods	method	NOUN
ijassa-372	20	3	like	like	ADP
ijassa-372	20	4	these	these	PRON
ijassa-372	20	5	,	,	PUNCT
ijassa-372	20	6	image	image	NOUN
ijassa-372	20	7	is	be	AUX
ijassa-372	20	8	divided	divide	VERB
ijassa-372	20	9	into	into	ADP
ijassa-372	20	10	several	several	ADJ
ijassa-372	20	11	non	non	ADJ
ijassa-372	20	12	-	-	ADJ
ijassa-372	20	13	overlapping	overlap	VERB
ijassa-372	20	14	rectangular	rectangular	ADJ
ijassa-372	20	15	patches	patch	NOUN
ijassa-372	20	16	,	,	PUNCT
ijassa-372	20	17	and	and	CCONJ
ijassa-372	20	18	the	the	DET
ijassa-372	20	19	feature	feature	NOUN
ijassa-372	20	20	histogram	histogram	NOUN
ijassa-372	20	21	of	of	ADP
ijassa-372	20	22	each	each	DET
ijassa-372	20	23	patches	patch	NOUN
ijassa-372	20	24	is	be	AUX
ijassa-372	20	25	computed	compute	VERB
ijassa-372	20	26	and	and	CCONJ
ijassa-372	20	27	combined	combine	VERB
ijassa-372	20	28	as	as	ADP
ijassa-372	20	29	the	the	DET
ijassa-372	20	30	image	image	NOUN
ijassa-372	20	31	feature	feature	NOUN
ijassa-372	20	32	.	.	PUNCT
ijassa-372	21	1	consequently	consequently	ADV
ijassa-372	21	2	,	,	PUNCT
ijassa-372	21	3	the	the	DET
ijassa-372	21	4	distance	distance	NOUN
ijassa-372	21	5	between	between	ADP
ijassa-372	21	6	images	image	NOUN
ijassa-372	21	7	is	be	AUX
ijassa-372	21	8	the	the	DET
ijassa-372	21	9	weighted	weighted	ADJ
ijassa-372	21	10	sum	sum	NOUN
ijassa-372	21	11	of	of	ADP
ijassa-372	21	12	distances	distance	NOUN
ijassa-372	21	13	between	between	ADP
ijassa-372	21	14	corresponding	correspond	VERB
ijassa-372	21	15	patches	patch	NOUN
ijassa-372	21	16	locating	locate	VERB
ijassa-372	21	17	in	in	ADP
ijassa-372	21	18	the	the	DET
ijassa-372	21	19	same	same	ADJ
ijassa-372	21	20	spatial	spatial	ADJ
ijassa-372	21	21	positions	position	NOUN
ijassa-372	21	22	.	.	PUNCT
ijassa-372	22	1	but	but	CCONJ
ijassa-372	22	2	we	we	PRON
ijassa-372	22	3	know	know	VERB
ijassa-372	22	4	,	,	PUNCT
ijassa-372	22	5	the	the	DET
ijassa-372	22	6	foreground	foreground	NOUN
ijassa-372	22	7	objects	object	NOUN
ijassa-372	22	8	of	of	ADP
ijassa-372	22	9	similar	similar	ADJ
ijassa-372	22	10	images	image	NOUN
ijassa-372	22	11	maybe	maybe	ADV
ijassa-372	22	12	lie	lie	VERB
ijassa-372	22	13	in	in	ADP
ijassa-372	22	14	different	different	ADJ
ijassa-372	22	15	spatial	spatial	ADJ
ijassa-372	22	16	locations	location	NOUN
ijassa-372	22	17	,	,	PUNCT
ijassa-372	22	18	poses	pose	NOUN
ijassa-372	22	19	and	and	CCONJ
ijassa-372	22	20	orientations	orientation	NOUN
ijassa-372	22	21	,	,	PUNCT
ijassa-372	22	22	which	which	PRON
ijassa-372	22	23	make	make	VERB
ijassa-372	22	24	the	the	DET
ijassa-372	22	25	distance	distance	NOUN
ijassa-372	22	26	measure	measure	NOUN
ijassa-372	22	27	is	be	AUX
ijassa-372	22	28	not	not	PART
ijassa-372	22	29	accurate	accurate	ADJ
ijassa-372	22	30	in	in	ADP
ijassa-372	22	31	many	many	ADJ
ijassa-372	22	32	situations	situation	NOUN
ijassa-372	22	33	.	.	PUNCT
ijassa-372	23	1	in	in	ADP
ijassa-372	23	2	this	this	DET
ijassa-372	23	3	work	work	NOUN
ijassa-372	23	4	,	,	PUNCT
ijassa-372	23	5	we	we	PRON
ijassa-372	23	6	present	present	VERB
ijassa-372	23	7	a	a	DET
ijassa-372	23	8	more	more	ADV
ijassa-372	23	9	flexible	flexible	ADJ
ijassa-372	23	10	patch	patch	NOUN
ijassa-372	23	11	-	-	PUNCT
ijassa-372	23	12	based	base	VERB
ijassa-372	23	13	distance	distance	NOUN
ijassa-372	23	14	measure	measure	NOUN
ijassa-372	23	15	method	method	NOUN
ijassa-372	23	16	integrating	integrate	VERB
ijassa-372	23	17	with	with	ADP
ijassa-372	23	18	bow	bow	NOUN
ijassa-372	23	19	,	,	PUNCT
ijassa-372	23	20	aiming	aim	VERB
ijassa-372	23	21	at	at	ADP
ijassa-372	23	22	increasing	increase	VERB
ijassa-372	23	23	the	the	DET
ijassa-372	23	24	precision	precision	NOUN
ijassa-372	23	25	of	of	ADP
ijassa-372	23	26	the	the	DET
ijassa-372	23	27	traditional	traditional	ADJ
ijassa-372	23	28	bow	bow	NOUN
ijassa-372	23	29	for	for	ADP
ijassa-372	23	30	semantic	semantic	ADJ
ijassa-372	23	31	retrieval	retrieval	NOUN
ijassa-372	23	32	.	.	PUNCT
ijassa-372	24	1	each	each	DET
ijassa-372	24	2	patch	patch	NOUN
ijassa-372	24	3	of	of	ADP
ijassa-372	24	4	the	the	DET
ijassa-372	24	5	first	first	ADJ
ijassa-372	24	6	image	image	NOUN
ijassa-372	24	7	is	be	AUX
ijassa-372	24	8	assigned	assign	VERB
ijassa-372	24	9	to	to	ADP
ijassa-372	24	10	exactly	exactly	ADV
ijassa-372	24	11	one	one	NUM
ijassa-372	24	12	patch	patch	NOUN
ijassa-372	24	13	of	of	ADP
ijassa-372	24	14	the	the	DET
ijassa-372	24	15	second	second	ADJ
ijassa-372	24	16	one	one	NOUN
ijassa-372	24	17	considering	consider	VERB
ijassa-372	24	18	both	both	CCONJ
ijassa-372	24	19	the	the	DET
ijassa-372	24	20	bow	bow	NOUN
ijassa-372	24	21	feature	feature	NOUN
ijassa-372	24	22	and	and	CCONJ
ijassa-372	24	23	the	the	DET
ijassa-372	24	24	spatial	spatial	ADJ
ijassa-372	24	25	location	location	NOUN
ijassa-372	24	26	of	of	ADP
ijassa-372	24	27	patches	patch	NOUN
ijassa-372	24	28	in	in	ADP
ijassa-372	24	29	the	the	DET
ijassa-372	24	30	image	image	NOUN
ijassa-372	24	31	,	,	PUNCT
ijassa-372	24	32	and	and	CCONJ
ijassa-372	24	33	the	the	DET
ijassa-372	24	34	distance	distance	NOUN
ijassa-372	24	35	measure	measure	NOUN
ijassa-372	24	36	problem	problem	NOUN
ijassa-372	24	37	of	of	ADP
ijassa-372	24	38	images	image	NOUN
ijassa-372	24	39	is	be	AUX
ijassa-372	24	40	transformed	transform	VERB
ijassa-372	24	41	into	into	ADP
ijassa-372	24	42	square	square	ADJ
ijassa-372	24	43	assignment	assignment	NOUN
ijassa-372	24	44	problem	problem	NOUN
ijassa-372	24	45	.	.	PUNCT
ijassa-372	25	1	our	our	PRON
ijassa-372	25	2	idea	idea	NOUN
ijassa-372	25	3	is	be	AUX
ijassa-372	25	4	similar	similar	ADJ
ijassa-372	25	5	to	to	ADP
ijassa-372	25	6	[	[	X
ijassa-372	25	7	7	7	X
ijassa-372	25	8	]	]	PUNCT
ijassa-372	25	9	which	which	PRON
ijassa-372	25	10	works	work	VERB
ijassa-372	25	11	in	in	ADP
ijassa-372	25	12	pixel	pixel	PROPN
ijassa-372	25	13	level	level	NOUN
ijassa-372	25	14	and	and	CCONJ
ijassa-372	25	15	only	only	ADV
ijassa-372	25	16	makes	make	VERB
ijassa-372	25	17	use	use	NOUN
ijassa-372	25	18	of	of	ADP
ijassa-372	25	19	image	image	NOUN
ijassa-372	25	20	color	color	NOUN
ijassa-372	25	21	feature	feature	NOUN
ijassa-372	25	22	.	.	PUNCT
ijassa-372	26	1	however	however	ADV
ijassa-372	26	2	,	,	PUNCT
ijassa-372	26	3	our	our	PRON
ijassa-372	26	4	method	method	NOUN
ijassa-372	26	5	works	work	VERB
ijassa-372	26	6	in	in	ADP
ijassa-372	26	7	patch	patch	ADJ
ijassa-372	26	8	level	level	NOUN
ijassa-372	26	9	and	and	CCONJ
ijassa-372	26	10	utilize	utilize	VERB
ijassa-372	26	11	image	image	NOUN
ijassa-372	26	12	local	local	ADJ
ijassa-372	26	13	features	feature	NOUN
ijassa-372	26	14	which	which	PRON
ijassa-372	26	15	have	have	VERB
ijassa-372	26	16	stronger	strong	ADJ
ijassa-372	26	17	descriptive	descriptive	ADJ
ijassa-372	26	18	power	power	NOUN
ijassa-372	26	19	.	.	PUNCT
ijassa-372	27	1	the	the	DET
ijassa-372	27	2	rest	rest	NOUN
ijassa-372	27	3	of	of	ADP
ijassa-372	27	4	this	this	DET
ijassa-372	27	5	paper	paper	NOUN
ijassa-372	27	6	is	be	AUX
ijassa-372	27	7	organized	organize	VERB
ijassa-372	27	8	as	as	SCONJ
ijassa-372	27	9	follows	follow	VERB
ijassa-372	27	10	.	.	PUNCT
ijassa-372	28	1	section	section	NOUN
ijassa-372	28	2	2	2	NUM
ijassa-372	28	3	discusses	discuss	VERB
ijassa-372	28	4	the	the	DET
ijassa-372	28	5	related	related	ADJ
ijassa-372	28	6	work	work	NOUN
ijassa-372	28	7	.	.	PUNCT
ijassa-372	29	1	section	section	NOUN
ijassa-372	29	2	3	3	NUM
ijassa-372	29	3	introduces	introduce	VERB
ijassa-372	29	4	the	the	DET
ijassa-372	29	5	image	image	NOUN
ijassa-372	29	6	feature	feature	NOUN
ijassa-372	29	7	representation	representation	NOUN
ijassa-372	29	8	as	as	ADP
ijassa-372	29	9	a	a	DET
ijassa-372	29	10	set	set	NOUN
ijassa-372	29	11	of	of	ADP
ijassa-372	29	12	patches	patch	NOUN
ijassa-372	29	13	.	.	PUNCT
ijassa-372	30	1	section	section	NOUN
ijassa-372	30	2	4	4	NUM
ijassa-372	30	3	details	detail	NOUN
ijassa-372	30	4	the	the	DET
ijassa-372	30	5	proposed	propose	VERB
ijassa-372	30	6	image	image	NOUN
ijassa-372	30	7	distance	distance	NOUN
ijassa-372	30	8	measure	measure	NOUN
ijassa-372	30	9	.	.	PUNCT
ijassa-372	31	1	section	section	NOUN
ijassa-372	31	2	5	5	NUM
ijassa-372	31	3	gives	give	VERB
ijassa-372	31	4	the	the	DET
ijassa-372	31	5	experimental	experimental	ADJ
ijassa-372	31	6	results	result	NOUN
ijassa-372	31	7	on	on	ADP
ijassa-372	31	8	image	image	NOUN
ijassa-372	31	9	retrieval	retrieval	NOUN
ijassa-372	31	10	application	application	NOUN
ijassa-372	31	11	.	.	PUNCT
ijassa-372	32	1	section	section	NOUN
ijassa-372	32	2	6	6	NUM
ijassa-372	32	3	concludes	conclude	VERB
ijassa-372	32	4	the	the	DET
ijassa-372	32	5	paper	paper	NOUN
ijassa-372	32	6	.	.	PUNCT
ijassa-372	33	1	2	2	NUM
ijassa-372	33	2	related	relate	VERB
ijassa-372	33	3	work	work	NOUN
ijassa-372	33	4	there	there	PRON
ijassa-372	33	5	are	be	VERB
ijassa-372	33	6	several	several	ADJ
ijassa-372	33	7	recent	recent	ADJ
ijassa-372	33	8	works	work	NOUN
ijassa-372	33	9	in	in	ADP
ijassa-372	33	10	the	the	DET
ijassa-372	33	11	content	content	NOUN
ijassa-372	33	12	-	-	PUNCT
ijassa-372	33	13	based	base	VERB
ijassa-372	33	14	image	image	NOUN
ijassa-372	33	15	retrieval	retrieval	NOUN
ijassa-372	33	16	(	(	PUNCT
ijassa-372	33	17	cbir	cbir	NOUN
ijassa-372	33	18	)	)	PUNCT
ijassa-372	33	19	area	area	NOUN
ijassa-372	33	20	which	which	PRON
ijassa-372	33	21	present	present	VERB
ijassa-372	33	22	interesting	interesting	ADJ
ijassa-372	33	23	advances	advance	NOUN
ijassa-372	33	24	for	for	ADP
ijassa-372	33	25	encoding	encode	VERB
ijassa-372	33	26	spatial	spatial	ADJ
ijassa-372	33	27	information	information	NOUN
ijassa-372	33	28	of	of	ADP
ijassa-372	33	29	visual	visual	ADJ
ijassa-372	33	30	words	word	NOUN
ijassa-372	33	31	.	.	PUNCT
ijassa-372	34	1	in	in	ADP
ijassa-372	34	2	the	the	DET
ijassa-372	34	3	early	early	ADJ
ijassa-372	34	4	days	day	NOUN
ijassa-372	34	5	,	,	PUNCT
ijassa-372	34	6	researchers	researcher	NOUN
ijassa-372	34	7	present	present	VERB
ijassa-372	34	8	color	color	NOUN
ijassa-372	34	9	correlogram	correlogram	NOUN
ijassa-372	35	1	[	[	X
ijassa-372	35	2	8	8	NUM
ijassa-372	35	3	]	]	PUNCT
ijassa-372	35	4	to	to	PART
ijassa-372	35	5	encode	encode	VERB
ijassa-372	35	6	the	the	DET
ijassa-372	35	7	spatial	spatial	ADJ
ijassa-372	35	8	arrangement	arrangement	NOUN
ijassa-372	35	9	of	of	ADP
ijassa-372	35	10	colors	color	NOUN
ijassa-372	35	11	.	.	PUNCT
ijassa-372	36	1	now	now	ADV
ijassa-372	36	2	the	the	DET
ijassa-372	36	3	study	study	NOUN
ijassa-372	36	4	object	object	NOUN
ijassa-372	36	5	is	be	AUX
ijassa-372	36	6	turned	turn	VERB
ijassa-372	36	7	into	into	ADP
ijassa-372	36	8	local	local	ADJ
ijassa-372	36	9	feature	feature	NOUN
ijassa-372	36	10	rather	rather	ADV
ijassa-372	36	11	than	than	ADP
ijassa-372	36	12	pixel	pixel	PROPN
ijassa-372	36	13	color	color	NOUN
ijassa-372	36	14	.	.	PUNCT
ijassa-372	37	1	the	the	DET
ijassa-372	37	2	most	most	ADV
ijassa-372	37	3	popular	popular	ADJ
ijassa-372	37	4	approach	approach	NOUN
ijassa-372	37	5	to	to	PART
ijassa-372	37	6	encode	encode	VERB
ijassa-372	37	7	the	the	DET
ijassa-372	37	8	spatial	spatial	ADJ
ijassa-372	37	9	information	information	NOUN
ijassa-372	37	10	of	of	ADP
ijassa-372	37	11	visual	visual	ADJ
ijassa-372	37	12	words	word	NOUN
ijassa-372	37	13	is	be	AUX
ijassa-372	37	14	spatial	spatial	ADJ
ijassa-372	37	15	pyramid	pyramid	NOUN
ijassa-372	37	16	matching	matching	NOUN
ijassa-372	37	17	,	,	PUNCT
ijassa-372	37	18	which	which	PRON
ijassa-372	37	19	involves	involve	VERB
ijassa-372	37	20	repeatedly	repeatedly	ADV
ijassa-372	37	21	subdividing	subdivide	VERB
ijassa-372	37	22	the	the	DET
ijassa-372	37	23	image	image	NOUN
ijassa-372	37	24	and	and	CCONJ
ijassa-372	37	25	computing	compute	VERB
ijassa-372	37	26	histograms	histogram	NOUN
ijassa-372	37	27	of	of	ADP
ijassa-372	37	28	local	local	ADJ
ijassa-372	37	29	features	feature	NOUN
ijassa-372	37	30	at	at	ADP
ijassa-372	37	31	increasingly	increasingly	ADV
ijassa-372	37	32	fine	fine	ADJ
ijassa-372	37	33	resolutions	resolution	NOUN
ijassa-372	37	34	.	.	PUNCT
ijassa-372	38	1	it	it	PRON
ijassa-372	38	2	is	be	AUX
ijassa-372	38	3	succesful	succesful	ADJ
ijassa-372	38	4	for	for	ADP
ijassa-372	38	5	image	image	NOUN
ijassa-372	38	6	classification	classification	NOUN
ijassa-372	38	7	,	,	PUNCT
ijassa-372	38	8	however	however	ADV
ijassa-372	38	9	,	,	PUNCT
ijassa-372	38	10	it	it	PRON
ijassa-372	38	11	is	be	AUX
ijassa-372	38	12	not	not	PART
ijassa-372	38	13	robust	robust	ADJ
ijassa-372	38	14	for	for	ADP
ijassa-372	38	15	some	some	DET
ijassa-372	38	16	image	image	NOUN
ijassa-372	38	17	transformations	transformation	NOUN
ijassa-372	38	18	such	such	ADJ
ijassa-372	38	19	as	as	ADP
ijassa-372	38	20	roation	roation	NOUN
ijassa-372	38	21	.	.	PUNCT
ijassa-372	39	1	many	many	ADJ
ijassa-372	39	2	of	of	ADP
ijassa-372	39	3	the	the	DET
ijassa-372	39	4	existing	exist	VERB
ijassa-372	39	5	approaches	approach	NOUN
ijassa-372	39	6	leave	leave	VERB
ijassa-372	39	7	the	the	DET
ijassa-372	39	8	spatial	spatial	ADJ
ijassa-372	39	9	verification	verification	NOUN
ijassa-372	39	10	as	as	ADP
ijassa-372	39	11	a	a	DET
ijassa-372	39	12	post	post	ADJ
ijassa-372	39	13	-	-	ADJ
ijassa-372	39	14	processing	processing	ADJ
ijassa-372	39	15	step	step	NOUN
ijassa-372	39	16	.	.	PUNCT
ijassa-372	40	1	they	they	PRON
ijassa-372	40	2	find	find	VERB
ijassa-372	40	3	the	the	DET
ijassa-372	40	4	matching	matching	NOUN
ijassa-372	40	5	points	point	NOUN
ijassa-372	40	6	between	between	ADP
ijassa-372	40	7	images	image	NOUN
ijassa-372	40	8	,	,	PUNCT
ijassa-372	40	9	then	then	ADV
ijassa-372	40	10	compute	compute	VERB
ijassa-372	40	11	the	the	DET
ijassa-372	40	12	spatial	spatial	ADJ
ijassa-372	40	13	representation	representation	NOUN
ijassa-372	40	14	which	which	PRON
ijassa-372	40	15	is	be	AUX
ijassa-372	40	16	used	use	VERB
ijassa-372	40	17	to	to	PART
ijassa-372	40	18	identify	identify	VERB
ijassa-372	40	19	and	and	CCONJ
ijassa-372	40	20	remove	remove	VERB
ijassa-372	40	21	the	the	DET
ijassa-372	40	22	false	false	ADJ
ijassa-372	40	23	matches	match	NOUN
ijassa-372	40	24	,	,	PUNCT
ijassa-372	40	25	compute	compute	VERB
ijassa-372	40	26	the	the	DET
ijassa-372	40	27	similarity	similarity	NOUN
ijassa-372	40	28	between	between	ADP
ijassa-372	40	29	images	image	NOUN
ijassa-372	40	30	by	by	ADP
ijassa-372	40	31	using	use	VERB
ijassa-372	40	32	the	the	DET
ijassa-372	40	33	number	number	NOUN
ijassa-372	40	34	of	of	ADP
ijassa-372	40	35	matched	match	VERB
ijassa-372	40	36	features	feature	NOUN
ijassa-372	40	37	and	and	CCONJ
ijassa-372	40	38	a	a	DET
ijassa-372	40	39	penalty	penalty	NOUN
ijassa-372	40	40	term	term	NOUN
ijassa-372	40	41	from	from	ADP
ijassa-372	40	42	spatial	spatial	ADJ
ijassa-372	40	43	verification	verification	NOUN
ijassa-372	40	44	.	.	PUNCT
ijassa-372	41	1	ransac	ransac	NOUN
ijassa-372	41	2	exploits	exploit	VERB
ijassa-372	41	3	full	full	ADJ
ijassa-372	41	4	geometric	geometric	ADJ
ijassa-372	41	5	verification	verification	NOUN
ijassa-372	41	6	[	[	X
ijassa-372	41	7	9	9	NUM
ijassa-372	41	8	-	-	SYM
ijassa-372	41	9	10	10	NUM
ijassa-372	41	10	]	]	PUNCT
ijassa-372	41	11	,	,	PUNCT
ijassa-372	41	12	which	which	PRON
ijassa-372	41	13	can	can	AUX
ijassa-372	41	14	greatly	greatly	ADV
ijassa-372	41	15	improve	improve	VERB
ijassa-372	41	16	retrieval	retrieval	NOUN
ijassa-372	41	17	performance	performance	NOUN
ijassa-372	41	18	,	,	PUNCT
ijassa-372	41	19	but	but	CCONJ
ijassa-372	41	20	it	it	PRON
ijassa-372	41	21	is	be	AUX
ijassa-372	41	22	com244	com244	PROPN
ijassa-372	41	23	yan	yan	PROPN
ijassa-372	41	24	yu	yu	PROPN
ijassa-372	41	25	:	:	PUNCT
ijassa-372	41	26	image	image	NOUN
ijassa-372	41	27	retrieval	retrieval	NOUN
ijassa-372	41	28	based	base	VERB
ijassa-372	41	29	on	on	ADP
ijassa-372	41	30	spatial	spatial	ADJ
ijassa-372	41	31	organization	organization	NOUN
ijassa-372	41	32	of	of	ADP
ijassa-372	41	33	patches	patch	NOUN
ijassa-372	41	34	putationally	putationally	ADV
ijassa-372	41	35	expensive	expensive	ADJ
ijassa-372	41	36	.	.	PUNCT
ijassa-372	42	1	another	another	DET
ijassa-372	42	2	recent	recent	ADJ
ijassa-372	42	3	approach	approach	NOUN
ijassa-372	42	4	explicitly	explicitly	ADV
ijassa-372	42	5	encodes	encode	VERB
ijassa-372	42	6	the	the	DET
ijassa-372	42	7	spatial	spatial	ADJ
ijassa-372	42	8	relationship	relationship	NOUN
ijassa-372	42	9	among	among	ADP
ijassa-372	42	10	each	each	DET
ijassa-372	42	11	pair	pair	NOUN
ijassa-372	42	12	of	of	ADP
ijassa-372	42	13	matching	match	VERB
ijassa-372	42	14	points	point	NOUN
ijassa-372	42	15	by	by	ADP
ijassa-372	42	16	utilizing	utilize	VERB
ijassa-372	42	17	binary	binary	ADJ
ijassa-372	42	18	spatial	spatial	ADJ
ijassa-372	42	19	maps	map	NOUN
ijassa-372	42	20	[	[	X
ijassa-372	42	21	2	2	NUM
ijassa-372	42	22	]	]	PUNCT
ijassa-372	42	23	.	.	PUNCT
ijassa-372	43	1	this	this	DET
ijassa-372	43	2	method	method	NOUN
ijassa-372	43	3	is	be	AUX
ijassa-372	43	4	efficient	efficient	ADJ
ijassa-372	43	5	and	and	CCONJ
ijassa-372	43	6	effective	effective	ADJ
ijassa-372	43	7	,	,	PUNCT
ijassa-372	43	8	but	but	CCONJ
ijassa-372	43	9	its	its	PRON
ijassa-372	43	10	representation	representation	NOUN
ijassa-372	43	11	is	be	AUX
ijassa-372	43	12	very	very	ADV
ijassa-372	43	13	specific	specific	ADJ
ijassa-372	43	14	for	for	ADP
ijassa-372	43	15	partial	partial	ADJ
ijassa-372	43	16	-	-	PUNCT
ijassa-372	43	17	duplicate	duplicate	ADJ
ijassa-372	43	18	search	search	NOUN
ijassa-372	43	19	and	and	CCONJ
ijassa-372	43	20	it	it	PRON
ijassa-372	43	21	is	be	AUX
ijassa-372	43	22	not	not	PART
ijassa-372	43	23	suitable	suitable	ADJ
ijassa-372	43	24	for	for	ADP
ijassa-372	43	25	the	the	DET
ijassa-372	43	26	semantic	semantic	ADJ
ijassa-372	43	27	-	-	PUNCT
ijassa-372	43	28	search	search	NOUN
ijassa-372	43	29	application	application	NOUN
ijassa-372	43	30	.	.	PUNCT
ijassa-372	44	1	another	another	DET
ijassa-372	44	2	recent	recent	ADJ
ijassa-372	44	3	approah	approah	NOUN
ijassa-372	44	4	learns	learn	VERB
ijassa-372	44	5	the	the	DET
ijassa-372	44	6	postions	postion	NOUN
ijassa-372	44	7	of	of	ADP
ijassa-372	44	8	visual	visual	ADJ
ijassa-372	44	9	words	word	NOUN
ijassa-372	44	10	occurrences	occurrence	NOUN
ijassa-372	44	11	in	in	ADP
ijassa-372	44	12	an	an	DET
ijassa-372	44	13	image	image	NOUN
ijassa-372	44	14	dataset	dataset	NOUN
ijassa-372	44	15	,	,	PUNCT
ijassa-372	44	16	called	call	VERB
ijassa-372	44	17	lp	lp	ADJ
ijassa-372	44	18	-	-	PUNCT
ijassa-372	44	19	norm	norm	NOUN
ijassa-372	44	20	pooling	pooling	NOUN
ijassa-372	44	21	[	[	X
ijassa-372	44	22	11	11	NUM
ijassa-372	44	23	]	]	PUNCT
ijassa-372	44	24	.	.	PUNCT
ijassa-372	45	1	the	the	DET
ijassa-372	45	2	method	method	NOUN
ijassa-372	45	3	first	first	ADV
ijassa-372	45	4	transforms	transform	VERB
ijassa-372	45	5	all	all	DET
ijassa-372	45	6	the	the	DET
ijassa-372	45	7	images	image	NOUN
ijassa-372	45	8	into	into	ADP
ijassa-372	45	9	the	the	DET
ijassa-372	45	10	same	same	ADJ
ijassa-372	45	11	resolution	resolution	NOUN
ijassa-372	45	12	,	,	PUNCT
ijassa-372	45	13	and	and	CCONJ
ijassa-372	45	14	use	use	VERB
ijassa-372	45	15	a	a	DET
ijassa-372	45	16	dense	dense	ADJ
ijassa-372	45	17	sampling	sampling	NOUN
ijassa-372	45	18	which	which	PRON
ijassa-372	45	19	generates	generate	VERB
ijassa-372	45	20	m	m	NOUN
ijassa-372	45	21	points	point	NOUN
ijassa-372	45	22	in	in	ADP
ijassa-372	45	23	each	each	DET
ijassa-372	45	24	image	image	NOUN
ijassa-372	45	25	.	.	PUNCT
ijassa-372	46	1	each	each	DET
ijassa-372	46	2	visual	visual	ADJ
ijassa-372	46	3	word	word	NOUN
ijassa-372	46	4	w	w	NOUN
ijassa-372	46	5	has	have	VERB
ijassa-372	46	6	a	a	DET
ijassa-372	46	7	md	md	PROPN
ijassa-372	46	8	vector	vector	NOUN
ijassa-372	46	9	,	,	PUNCT
ijassa-372	46	10	in	in	ADP
ijassa-372	46	11	which	which	PRON
ijassa-372	46	12	the	the	DET
ijassa-372	46	13	m	m	NOUN
ijassa-372	46	14	-	-	PUNCT
ijassa-372	46	15	th	th	VERB
ijassa-372	46	16	value	value	NOUN
ijassa-372	46	17	representis	representis	NOUN
ijassa-372	46	18	the	the	DET
ijassa-372	46	19	m	m	PROPN
ijassa-372	46	20	-	-	PUNCT
ijassa-372	46	21	th	th	VERB
ijassa-372	46	22	point	point	NOUN
ijassa-372	46	23	activate	activate	VERB
ijassa-372	46	24	the	the	DET
ijassa-372	46	25	visual	visual	ADJ
ijassa-372	46	26	word	word	NOUN
ijassa-372	46	27	w.	w.	PROPN
ijassa-372	46	28	however	however	ADV
ijassa-372	46	29	,	,	PUNCT
ijassa-372	46	30	the	the	DET
ijassa-372	46	31	method	method	NOUN
ijassa-372	46	32	is	be	AUX
ijassa-372	46	33	designed	design	VERB
ijassa-372	46	34	for	for	ADP
ijassa-372	46	35	image	image	NOUN
ijassa-372	46	36	classification	classification	NOUN
ijassa-372	46	37	and	and	CCONJ
ijassa-372	46	38	it	it	PRON
ijassa-372	46	39	represent	represent	VERB
ijassa-372	46	40	the	the	DET
ijassa-372	46	41	absolute	absolute	ADJ
ijassa-372	46	42	postions	postion	NOUN
ijassa-372	46	43	of	of	ADP
ijassa-372	46	44	visual	visual	ADJ
ijassa-372	46	45	words	word	NOUN
ijassa-372	46	46	,	,	PUNCT
ijassa-372	46	47	which	which	PRON
ijassa-372	46	48	is	be	AUX
ijassa-372	46	49	not	not	PART
ijassa-372	46	50	robust	robust	ADJ
ijassa-372	46	51	to	to	ADP
ijassa-372	46	52	translation	translation	NOUN
ijassa-372	46	53	or	or	CCONJ
ijassa-372	46	54	rotation	rotation	NOUN
ijassa-372	46	55	of	of	ADP
ijassa-372	46	56	the	the	DET
ijassa-372	46	57	object	object	NOUN
ijassa-372	46	58	in	in	ADP
ijassa-372	46	59	the	the	DET
ijassa-372	46	60	image	image	NOUN
ijassa-372	46	61	.	.	PUNCT
ijassa-372	47	1	a	a	DET
ijassa-372	47	2	recent	recent	ADJ
ijassa-372	47	3	image	image	NOUN
ijassa-372	47	4	content	content	NOUN
ijassa-372	47	5	description	description	NOUN
ijassa-372	47	6	is	be	AUX
ijassa-372	47	7	proposed	propose	VERB
ijassa-372	47	8	for	for	ADP
ijassa-372	47	9	describing	describe	VERB
ijassa-372	47	10	the	the	DET
ijassa-372	47	11	spatial	spatial	ADJ
ijassa-372	47	12	layout	layout	NOUN
ijassa-372	47	13	of	of	ADP
ijassa-372	47	14	visual	visual	ADJ
ijassa-372	47	15	words	word	NOUN
ijassa-372	47	16	,	,	PUNCT
ijassa-372	47	17	called	call	VERB
ijassa-372	47	18	∆−tsr	∆−tsr	NOUN
ijassa-372	47	19	[	[	X
ijassa-372	47	20	12	12	NUM
ijassa-372	47	21	]	]	PUNCT
ijassa-372	47	22	.	.	PUNCT
ijassa-372	48	1	triangle	triangle	NOUN
ijassa-372	48	2	is	be	AUX
ijassa-372	48	3	used	use	VERB
ijassa-372	48	4	to	to	PART
ijassa-372	48	5	describe	describe	VERB
ijassa-372	48	6	the	the	DET
ijassa-372	48	7	relationship	relationship	NOUN
ijassa-372	48	8	among	among	ADP
ijassa-372	48	9	arbitrary	arbitrary	ADJ
ijassa-372	48	10	three	three	NUM
ijassa-372	48	11	interesting	interesting	ADJ
ijassa-372	48	12	points	point	NOUN
ijassa-372	48	13	,	,	PUNCT
ijassa-372	48	14	which	which	PRON
ijassa-372	48	15	is	be	AUX
ijassa-372	48	16	represented	represent	VERB
ijassa-372	48	17	as	as	ADP
ijassa-372	48	18	a	a	DET
ijassa-372	48	19	7d	7d	NUM
ijassa-372	48	20	signature	signature	NOUN
ijassa-372	48	21	composed	compose	VERB
ijassa-372	48	22	of	of	ADP
ijassa-372	48	23	labels	label	NOUN
ijassa-372	48	24	of	of	ADP
ijassa-372	48	25	correspoind	correspoind	ADJ
ijassa-372	48	26	viusal	viusal	ADJ
ijassa-372	48	27	words	word	NOUN
ijassa-372	48	28	,	,	PUNCT
ijassa-372	48	29	angles	angle	NOUN
ijassa-372	48	30	of	of	ADP
ijassa-372	48	31	vertices	vertex	NOUN
ijassa-372	48	32	,	,	PUNCT
ijassa-372	48	33	relative	relative	ADJ
ijassa-372	48	34	orientations	orientation	NOUN
ijassa-372	48	35	and	and	CCONJ
ijassa-372	48	36	scales	scale	NOUN
ijassa-372	48	37	of	of	ADP
ijassa-372	48	38	points	point	NOUN
ijassa-372	48	39	.	.	PUNCT
ijassa-372	49	1	after	after	SCONJ
ijassa-372	49	2	the	the	DET
ijassa-372	49	3	strategy	strategy	NOUN
ijassa-372	49	4	of	of	ADP
ijassa-372	49	5	triangle	triangle	NOUN
ijassa-372	49	6	selection	selection	NOUN
ijassa-372	49	7	is	be	AUX
ijassa-372	49	8	applied	apply	VERB
ijassa-372	49	9	,	,	PUNCT
ijassa-372	49	10	the	the	DET
ijassa-372	49	11	similarity	similarity	NOUN
ijassa-372	49	12	between	between	ADP
ijassa-372	49	13	images	image	NOUN
ijassa-372	49	14	is	be	AUX
ijassa-372	49	15	defined	define	VERB
ijassa-372	49	16	as	as	ADP
ijassa-372	49	17	the	the	DET
ijassa-372	49	18	ratio	ratio	NOUN
ijassa-372	49	19	of	of	ADP
ijassa-372	49	20	the	the	DET
ijassa-372	49	21	sum	sum	NOUN
ijassa-372	49	22	of	of	ADP
ijassa-372	49	23	similairity	similairity	NOUN
ijassa-372	49	24	between	between	ADP
ijassa-372	49	25	the	the	DET
ijassa-372	49	26	most	most	ADV
ijassa-372	49	27	similar	similar	ADJ
ijassa-372	49	28	triangles	triangle	NOUN
ijassa-372	49	29	and	and	CCONJ
ijassa-372	49	30	the	the	DET
ijassa-372	49	31	cardinality	cardinality	NOUN
ijassa-372	49	32	of	of	ADP
ijassa-372	49	33	the	the	DET
ijassa-372	49	34	set	set	NOUN
ijassa-372	49	35	of	of	ADP
ijassa-372	49	36	triangles	triangle	NOUN
ijassa-372	49	37	.	.	PUNCT
ijassa-372	50	1	this	this	DET
ijassa-372	50	2	method	method	NOUN
ijassa-372	50	3	explicitly	explicitly	ADV
ijassa-372	50	4	encodes	encode	VERB
ijassa-372	50	5	the	the	DET
ijassa-372	50	6	spatial	spatial	ADJ
ijassa-372	50	7	relationship	relationship	NOUN
ijassa-372	50	8	among	among	ADP
ijassa-372	50	9	visual	visual	ADJ
ijassa-372	50	10	words	word	NOUN
ijassa-372	50	11	,	,	PUNCT
ijassa-372	50	12	but	but	CCONJ
ijassa-372	50	13	it	it	PRON
ijassa-372	50	14	only	only	ADV
ijassa-372	50	15	works	work	VERB
ijassa-372	50	16	with	with	ADP
ijassa-372	50	17	hard	hard	ADJ
ijassa-372	50	18	assignment	assignment	NOUN
ijassa-372	50	19	.	.	PUNCT
ijassa-372	51	1	more	more	ADV
ijassa-372	51	2	recently	recently	ADV
ijassa-372	51	3	,	,	PUNCT
ijassa-372	51	4	graph	graph	NOUN
ijassa-372	51	5	-	-	PUNCT
ijassa-372	51	6	based	base	VERB
ijassa-372	51	7	approaches	approach	NOUN
ijassa-372	51	8	for	for	ADP
ijassa-372	51	9	encoding	encode	VERB
ijassa-372	51	10	spatial	spatial	ADJ
ijassa-372	51	11	information	information	NOUN
ijassa-372	51	12	are	be	AUX
ijassa-372	51	13	proposed	propose	VERB
ijassa-372	51	14	.	.	PUNCT
ijassa-372	52	1	the	the	DET
ijassa-372	52	2	nested	nested	ADJ
ijassa-372	52	3	multi	multi	ADJ
ijassa-372	52	4	-	-	ADJ
ijassa-372	52	5	layered	layered	ADJ
ijassa-372	52	6	local	local	ADJ
ijassa-372	52	7	graphs	graph	NOUN
ijassa-372	52	8	[	[	X
ijassa-372	52	9	13	13	NUM
ijassa-372	52	10	]	]	PUNCT
ijassa-372	52	11	is	be	AUX
ijassa-372	52	12	proposed	propose	VERB
ijassa-372	52	13	by	by	ADP
ijassa-372	52	14	building	build	VERB
ijassa-372	52	15	upon	upon	SCONJ
ijassa-372	52	16	sets	set	NOUN
ijassa-372	52	17	of	of	ADP
ijassa-372	52	18	surf	surf	NOUN
ijassa-372	52	19	points	point	NOUN
ijassa-372	52	20	with	with	ADP
ijassa-372	52	21	delaunay	delaunay	PROPN
ijassa-372	52	22	triangulation	triangulation	NOUN
ijassa-372	52	23	.	.	PUNCT
ijassa-372	53	1	a	a	DET
ijassa-372	53	2	bow	bow	NOUN
ijassa-372	53	3	framework	framework	NOUN
ijassa-372	53	4	is	be	AUX
ijassa-372	53	5	applied	apply	VERB
ijassa-372	53	6	on	on	ADP
ijassa-372	53	7	these	these	DET
ijassa-372	53	8	graphs	graph	NOUN
ijassa-372	53	9	,	,	PUNCT
ijassa-372	53	10	giving	give	VERB
ijassa-372	53	11	birth	birth	NOUN
ijassa-372	53	12	to	to	ADP
ijassa-372	53	13	a	a	DET
ijassa-372	53	14	bag	bag	NOUN
ijassa-372	53	15	-	-	PUNCT
ijassa-372	53	16	of	of	ADP
ijassa-372	53	17	-	-	PUNCT
ijassa-372	53	18	graphwords	graphword	NOUN
ijassa-372	53	19	representation	representation	NOUN
ijassa-372	53	20	,	,	PUNCT
ijassa-372	53	21	visual	visual	ADJ
ijassa-372	53	22	dictionary	dictionary	NOUN
ijassa-372	53	23	is	be	AUX
ijassa-372	53	24	built	build	VERB
ijassa-372	53	25	for	for	ADP
ijassa-372	53	26	each	each	DET
ijassa-372	53	27	layer	layer	NOUN
ijassa-372	53	28	of	of	ADP
ijassa-372	53	29	graphs	graph	NOUN
ijassa-372	53	30	and	and	CCONJ
ijassa-372	53	31	l1	l1	PROPN
ijassa-372	53	32	distance	distance	NOUN
ijassa-372	53	33	function	function	NOUN
ijassa-372	53	34	is	be	AUX
ijassa-372	53	35	used	use	VERB
ijassa-372	53	36	for	for	ADP
ijassa-372	53	37	image	image	NOUN
ijassa-372	53	38	retrieval	retrieval	NOUN
ijassa-372	53	39	.	.	PUNCT
ijassa-372	54	1	this	this	DET
ijassa-372	54	2	method	method	NOUN
ijassa-372	54	3	explicitly	explicitly	ADV
ijassa-372	54	4	encodes	encode	VERB
ijassa-372	54	5	the	the	DET
ijassa-372	54	6	spatial	spatial	ADJ
ijassa-372	54	7	relationship	relationship	NOUN
ijassa-372	54	8	of	of	ADP
ijassa-372	54	9	image	image	NOUN
ijassa-372	54	10	points	point	NOUN
ijassa-372	54	11	,	,	PUNCT
ijassa-372	54	12	however	however	ADV
ijassa-372	54	13	it	it	PRON
ijassa-372	54	14	has	have	VERB
ijassa-372	54	15	higher	high	ADJ
ijassa-372	54	16	computational	computational	ADJ
ijassa-372	54	17	cost	cost	NOUN
ijassa-372	54	18	and	and	CCONJ
ijassa-372	54	19	only	only	ADV
ijassa-372	54	20	work	work	VERB
ijassa-372	54	21	with	with	ADP
ijassa-372	54	22	hard	hard	ADJ
ijassa-372	54	23	assignment	assignment	NOUN
ijassa-372	54	24	.	.	PUNCT
ijassa-372	55	1	in	in	ADP
ijassa-372	55	2	this	this	DET
ijassa-372	55	3	work	work	NOUN
ijassa-372	55	4	,	,	PUNCT
ijassa-372	55	5	the	the	DET
ijassa-372	55	6	content	content	NOUN
ijassa-372	55	7	representation	representation	NOUN
ijassa-372	55	8	and	and	CCONJ
ijassa-372	55	9	similarity	similarity	NOUN
ijassa-372	55	10	measure	measure	NOUN
ijassa-372	55	11	method	method	NOUN
ijassa-372	55	12	of	of	ADP
ijassa-372	55	13	images	image	NOUN
ijassa-372	55	14	can	can	AUX
ijassa-372	55	15	work	work	VERB
ijassa-372	55	16	with	with	ADP
ijassa-372	55	17	both	both	CCONJ
ijassa-372	55	18	hard	hard	ADJ
ijassa-372	55	19	assignment	assignment	NOUN
ijassa-372	55	20	and	and	CCONJ
ijassa-372	55	21	soft	soft	ADJ
ijassa-372	55	22	assignment	assignment	NOUN
ijassa-372	55	23	,	,	PUNCT
ijassa-372	55	24	furthermoe	furthermoe	PROPN
ijassa-372	55	25	,	,	PUNCT
ijassa-372	55	26	it	it	PRON
ijassa-372	55	27	is	be	AUX
ijassa-372	55	28	suitable	suitable	ADJ
ijassa-372	55	29	for	for	ADP
ijassa-372	55	30	semantic	semantic	ADJ
ijassa-372	55	31	-	-	PUNCT
ijassa-372	55	32	retrieval	retrieval	NOUN
ijassa-372	55	33	.	.	PUNCT
ijassa-372	56	1	3	3	NUM
ijassa-372	56	2	proposed	propose	VERB
ijassa-372	56	3	method	method	NOUN
ijassa-372	56	4	3.1	3.1	NUM
ijassa-372	56	5	image	image	NOUN
ijassa-372	56	6	as	as	ADP
ijassa-372	56	7	a	a	DET
ijassa-372	56	8	set	set	NOUN
ijassa-372	56	9	of	of	ADP
ijassa-372	56	10	patches	patch	NOUN
ijassa-372	56	11	for	for	ADP
ijassa-372	56	12	a	a	DET
ijassa-372	56	13	given	give	VERB
ijassa-372	56	14	image	image	NOUN
ijassa-372	56	15	,	,	PUNCT
ijassa-372	56	16	the	the	DET
ijassa-372	56	17	set	set	NOUN
ijassa-372	56	18	of	of	ADP
ijassa-372	56	19	patches	patch	NOUN
ijassa-372	56	20	can	can	AUX
ijassa-372	56	21	be	be	AUX
ijassa-372	56	22	represented	represent	VERB
ijassa-372	56	23	as	as	ADP
ijassa-372	56	24	{	{	PUNCT
ijassa-372	56	25	fi}i=1,	fi}i=1,	NOUN
ijassa-372	56	26	...	...	PUNCT
ijassa-372	56	27	,n	,n	NOUN
ijassa-372	56	28	,	,	PUNCT
ijassa-372	56	29	fi	fi	NOUN
ijassa-372	56	30	∈	∈	PROPN
ijassa-372	56	31	rm	rm	NOUN
ijassa-372	56	32	(	(	PUNCT
ijassa-372	56	33	1	1	NUM
ijassa-372	56	34	)	)	PUNCT
ijassa-372	56	35	where	where	SCONJ
ijassa-372	56	36	fi	fi	NOUN
ijassa-372	56	37	is	be	AUX
ijassa-372	56	38	the	the	DET
ijassa-372	56	39	feature	feature	NOUN
ijassa-372	56	40	vector	vector	NOUN
ijassa-372	56	41	in	in	ADP
ijassa-372	56	42	the	the	DET
ijassa-372	56	43	i	i	PROPN
ijassa-372	56	44	-	-	PUNCT
ijassa-372	56	45	th	th	VERB
ijassa-372	56	46	image	image	NOUN
ijassa-372	56	47	patch	patch	NOUN
ijassa-372	56	48	,	,	PUNCT
ijassa-372	56	49	n	n	X
ijassa-372	56	50	is	be	AUX
ijassa-372	56	51	the	the	DET
ijassa-372	56	52	total	total	ADJ
ijassa-372	56	53	number	number	NOUN
ijassa-372	56	54	of	of	ADP
ijassa-372	56	55	image	image	NOUN
ijassa-372	56	56	patches	patch	NOUN
ijassa-372	56	57	and	and	CCONJ
ijassa-372	56	58	m	m	NOUN
ijassa-372	56	59	is	be	AUX
ijassa-372	56	60	the	the	DET
ijassa-372	56	61	length	length	NOUN
ijassa-372	56	62	of	of	ADP
ijassa-372	56	63	the	the	DET
ijassa-372	56	64	feature	feature	NOUN
ijassa-372	56	65	vector	vector	NOUN
ijassa-372	56	66	.	.	PUNCT
ijassa-372	57	1	the	the	DET
ijassa-372	57	2	pixel	pixel	PROPN
ijassa-372	57	3	sizes	size	NOUN
ijassa-372	57	4	of	of	ADP
ijassa-372	57	5	images	image	NOUN
ijassa-372	57	6	in	in	ADP
ijassa-372	57	7	our	our	PRON
ijassa-372	57	8	dataset	dataset	NOUN
ijassa-372	57	9	are	be	AUX
ijassa-372	57	10	various	various	ADJ
ijassa-372	57	11	.	.	PUNCT
ijassa-372	58	1	in	in	ADP
ijassa-372	58	2	order	order	NOUN
ijassa-372	58	3	to	to	PART
ijassa-372	58	4	realize	realize	VERB
ijassa-372	58	5	the	the	DET
ijassa-372	58	6	assignment	assignment	NOUN
ijassa-372	58	7	relation	relation	NOUN
ijassa-372	58	8	between	between	ADP
ijassa-372	58	9	advances	advance	NOUN
ijassa-372	58	10	in	in	ADP
ijassa-372	58	11	systems	system	NOUN
ijassa-372	58	12	science	science	NOUN
ijassa-372	58	13	and	and	CCONJ
ijassa-372	58	14	application	application	NOUN
ijassa-372	58	15	(	(	PUNCT
ijassa-372	58	16	2015	2015	NUM
ijassa-372	58	17	)	)	PUNCT
ijassa-372	58	18	vol.15	vol.15	NOUN
ijassa-372	58	19	no.3	no.3	VERB
ijassa-372	58	20	245	245	NUM
ijassa-372	58	21	fig.1	fig.1	ADJ
ijassa-372	58	22	original	original	ADJ
ijassa-372	58	23	images	image	NOUN
ijassa-372	58	24	and	and	CCONJ
ijassa-372	58	25	their	their	PRON
ijassa-372	58	26	sets	set	NOUN
ijassa-372	58	27	of	of	ADP
ijassa-372	58	28	patches	patch	NOUN
ijassa-372	58	29	.	.	PUNCT
ijassa-372	59	1	patches	patch	NOUN
ijassa-372	59	2	of	of	ADP
ijassa-372	59	3	two	two	NUM
ijassa-372	59	4	images	image	NOUN
ijassa-372	59	5	,	,	PUNCT
ijassa-372	59	6	all	all	DET
ijassa-372	59	7	the	the	DET
ijassa-372	59	8	images	image	NOUN
ijassa-372	59	9	are	be	AUX
ijassa-372	59	10	divided	divide	VERB
ijassa-372	59	11	into	into	ADP
ijassa-372	59	12	patches	patch	NOUN
ijassa-372	59	13	of	of	ADP
ijassa-372	59	14	the	the	DET
ijassa-372	59	15	same	same	ADJ
ijassa-372	59	16	amount	amount	NOUN
ijassa-372	59	17	,	,	PUNCT
ijassa-372	59	18	rather	rather	ADV
ijassa-372	59	19	than	than	ADP
ijassa-372	59	20	limited	limit	VERB
ijassa-372	59	21	to	to	ADP
ijassa-372	59	22	the	the	DET
ijassa-372	59	23	same	same	ADJ
ijassa-372	59	24	size	size	NOUN
ijassa-372	59	25	of	of	ADP
ijassa-372	59	26	each	each	DET
ijassa-372	59	27	patch	patch	NOUN
ijassa-372	59	28	.	.	PUNCT
ijassa-372	60	1	for	for	ADP
ijassa-372	60	2	example	example	NOUN
ijassa-372	60	3	,	,	PUNCT
ijassa-372	60	4	figure1	figure1	PROPN
ijassa-372	60	5	shows	show	VERB
ijassa-372	60	6	that	that	SCONJ
ijassa-372	60	7	two	two	NUM
ijassa-372	60	8	images	image	NOUN
ijassa-372	60	9	are	be	AUX
ijassa-372	60	10	divided	divide	VERB
ijassa-372	60	11	into	into	ADP
ijassa-372	60	12	8	8	NUM
ijassa-372	60	13	rows	row	NOUN
ijassa-372	60	14	and	and	CCONJ
ijassa-372	60	15	8	8	NUM
ijassa-372	60	16	columns	column	NOUN
ijassa-372	60	17	respectively	respectively	ADV
ijassa-372	60	18	.	.	PUNCT
ijassa-372	61	1	although	although	SCONJ
ijassa-372	61	2	they	they	PRON
ijassa-372	61	3	have	have	VERB
ijassa-372	61	4	different	different	ADJ
ijassa-372	61	5	aspect	aspect	NOUN
ijassa-372	61	6	ratio	ratio	NOUN
ijassa-372	61	7	,	,	PUNCT
ijassa-372	61	8	they	they	PRON
ijassa-372	61	9	all	all	PRON
ijassa-372	61	10	have	have	VERB
ijassa-372	61	11	a	a	DET
ijassa-372	61	12	set	set	NOUN
ijassa-372	61	13	of	of	ADP
ijassa-372	61	14	64	64	NUM
ijassa-372	61	15	patches	patch	NOUN
ijassa-372	61	16	in	in	ADP
ijassa-372	61	17	total	total	NOUN
ijassa-372	61	18	.	.	PUNCT
ijassa-372	62	1	in	in	ADP
ijassa-372	62	2	our	our	PRON
ijassa-372	62	3	method	method	NOUN
ijassa-372	62	4	,	,	PUNCT
ijassa-372	62	5	we	we	PRON
ijassa-372	62	6	encode	encode	VERB
ijassa-372	62	7	the	the	DET
ijassa-372	62	8	appearance	appearance	NOUN
ijassa-372	62	9	feature	feature	NOUN
ijassa-372	62	10	for	for	ADP
ijassa-372	62	11	each	each	DET
ijassa-372	62	12	patch	patch	NOUN
ijassa-372	62	13	rather	rather	ADV
ijassa-372	62	14	than	than	ADP
ijassa-372	62	15	the	the	DET
ijassa-372	62	16	global	global	ADJ
ijassa-372	62	17	image	image	NOUN
ijassa-372	62	18	.	.	PUNCT
ijassa-372	63	1	the	the	DET
ijassa-372	63	2	feature	feature	NOUN
ijassa-372	63	3	vector	vector	NOUN
ijassa-372	63	4	of	of	ADP
ijassa-372	63	5	each	each	DET
ijassa-372	63	6	patch	patch	NOUN
ijassa-372	63	7	is	be	AUX
ijassa-372	63	8	extracted	extract	VERB
ijassa-372	63	9	via	via	ADP
ijassa-372	63	10	bow	bow	NOUN
ijassa-372	63	11	model	model	NOUN
ijassa-372	63	12	according	accord	VERB
ijassa-372	63	13	to	to	ADP
ijassa-372	63	14	the	the	DET
ijassa-372	63	15	following	follow	VERB
ijassa-372	63	16	steps	step	NOUN
ijassa-372	63	17	.	.	PUNCT
ijassa-372	64	1	firstly	firstly	ADV
ijassa-372	64	2	,	,	PUNCT
ijassa-372	64	3	the	the	DET
ijassa-372	64	4	images	image	NOUN
ijassa-372	64	5	were	be	AUX
ijassa-372	64	6	all	all	ADV
ijassa-372	64	7	preprocessed	preprocesse	VERB
ijassa-372	64	8	into	into	ADP
ijassa-372	64	9	gray	gray	ADJ
ijassa-372	64	10	scale	scale	NOUN
ijassa-372	64	11	and	and	CCONJ
ijassa-372	64	12	local	local	ADJ
ijassa-372	64	13	features	feature	NOUN
ijassa-372	64	14	are	be	AUX
ijassa-372	64	15	extracted	extract	VERB
ijassa-372	64	16	from	from	ADP
ijassa-372	64	17	images	image	NOUN
ijassa-372	64	18	in	in	ADP
ijassa-372	64	19	the	the	DET
ijassa-372	64	20	training	training	NOUN
ijassa-372	64	21	set	set	VERB
ijassa-372	64	22	by	by	ADP
ijassa-372	64	23	dense	dense	ADJ
ijassa-372	64	24	sampling	sampling	NOUN
ijassa-372	64	25	.	.	PUNCT
ijassa-372	65	1	in	in	ADP
ijassa-372	65	2	this	this	DET
ijassa-372	65	3	work	work	NOUN
ijassa-372	65	4	,	,	PUNCT
ijassa-372	65	5	sift	sift	VERB
ijassa-372	65	6	descriptor	descriptor	NOUN
ijassa-372	65	7	[	[	X
ijassa-372	65	8	14	14	NUM
ijassa-372	65	9	]	]	PUNCT
ijassa-372	65	10	is	be	AUX
ijassa-372	65	11	adopted	adopt	VERB
ijassa-372	65	12	as	as	ADP
ijassa-372	65	13	local	local	ADJ
ijassa-372	65	14	feature	feature	NOUN
ijassa-372	65	15	descriptor	descriptor	NOUN
ijassa-372	65	16	,	,	PUNCT
ijassa-372	65	17	because	because	SCONJ
ijassa-372	65	18	it	it	PRON
ijassa-372	65	19	has	have	VERB
ijassa-372	65	20	averagely	averagely	ADV
ijassa-372	65	21	the	the	DET
ijassa-372	65	22	best	good	ADJ
ijassa-372	65	23	performance	performance	NOUN
ijassa-372	65	24	among	among	ADP
ijassa-372	65	25	local	local	ADJ
ijassa-372	65	26	feature	feature	NOUN
ijassa-372	65	27	descriptors	descriptor	NOUN
ijassa-372	65	28	[	[	X
ijassa-372	65	29	15	15	NUM
ijassa-372	65	30	]	]	PUNCT
ijassa-372	65	31	.	.	PUNCT
ijassa-372	66	1	dense	dense	ADJ
ijassa-372	66	2	sampling	sampling	NOUN
ijassa-372	66	3	is	be	AUX
ijassa-372	66	4	realized	realize	VERB
ijassa-372	66	5	by	by	ADP
ijassa-372	66	6	a	a	DET
ijassa-372	66	7	sliding	slide	VERB
ijassa-372	66	8	window	window	NOUN
ijassa-372	66	9	in	in	ADP
ijassa-372	66	10	the	the	DET
ijassa-372	66	11	size	size	NOUN
ijassa-372	66	12	of	of	ADP
ijassa-372	66	13	16∗16	16∗16	NUM
ijassa-372	66	14	pixels	pixel	NOUN
ijassa-372	66	15	,	,	PUNCT
ijassa-372	66	16	which	which	PRON
ijassa-372	66	17	sliding	slide	VERB
ijassa-372	66	18	from	from	ADP
ijassa-372	66	19	top	top	NOUN
ijassa-372	66	20	to	to	AUX
ijassa-372	66	21	down	down	ADP
ijassa-372	66	22	and	and	CCONJ
ijassa-372	66	23	left	leave	VERB
ijassa-372	66	24	to	to	PART
ijassa-372	66	25	right	right	ADV
ijassa-372	66	26	in	in	ADP
ijassa-372	66	27	an	an	DET
ijassa-372	66	28	image	image	NOUN
ijassa-372	66	29	by	by	ADP
ijassa-372	66	30	a	a	DET
ijassa-372	66	31	step	step	NOUN
ijassa-372	66	32	of	of	ADP
ijassa-372	66	33	8	8	NUM
ijassa-372	66	34	pixels	pixel	NOUN
ijassa-372	66	35	.	.	PUNCT
ijassa-372	67	1	these	these	DET
ijassa-372	67	2	regions	region	NOUN
ijassa-372	67	3	are	be	AUX
ijassa-372	67	4	overlapping	overlap	VERB
ijassa-372	67	5	for	for	ADP
ijassa-372	67	6	obtaining	obtain	VERB
ijassa-372	67	7	enough	enough	ADJ
ijassa-372	67	8	abundant	abundant	ADJ
ijassa-372	67	9	local	local	ADJ
ijassa-372	67	10	features	feature	NOUN
ijassa-372	67	11	.	.	PUNCT
ijassa-372	68	1	secondly	secondly	ADV
ijassa-372	68	2	,	,	PUNCT
ijassa-372	68	3	all	all	DET
ijassa-372	68	4	the	the	DET
ijassa-372	68	5	feature	feature	NOUN
ijassa-372	68	6	vectors	vector	NOUN
ijassa-372	68	7	are	be	AUX
ijassa-372	68	8	quantized	quantize	VERB
ijassa-372	68	9	to	to	PART
ijassa-372	68	10	construct	construct	VERB
ijassa-372	68	11	a	a	DET
ijassa-372	68	12	visual	visual	ADJ
ijassa-372	68	13	dictionary	dictionary	NOUN
ijassa-372	68	14	by	by	ADP
ijassa-372	68	15	standard	standard	ADJ
ijassa-372	68	16	k	k	PROPN
ijassa-372	68	17	-	-	PUNCT
ijassa-372	68	18	means	mean	VERB
ijassa-372	68	19	clustering	cluster	VERB
ijassa-372	68	20	algorithm	algorithm	NOUN
ijassa-372	68	21	[	[	X
ijassa-372	68	22	16	16	NUM
ijassa-372	68	23	]	]	PUNCT
ijassa-372	68	24	.	.	PUNCT
ijassa-372	69	1	lastly	lastly	ADV
ijassa-372	69	2	,	,	PUNCT
ijassa-372	69	3	each	each	DET
ijassa-372	69	4	patch	patch	NOUN
ijassa-372	69	5	of	of	ADP
ijassa-372	69	6	each	each	DET
ijassa-372	69	7	image	image	NOUN
ijassa-372	69	8	is	be	AUX
ijassa-372	69	9	descripted	descripte	VERB
ijassa-372	69	10	as	as	ADP
ijassa-372	69	11	a	a	DET
ijassa-372	69	12	histogram	histogram	NOUN
ijassa-372	69	13	indicating	indicate	VERB
ijassa-372	69	14	the	the	DET
ijassa-372	69	15	probability	probability	NOUN
ijassa-372	69	16	distribution	distribution	NOUN
ijassa-372	69	17	of	of	ADP
ijassa-372	69	18	visual	visual	ADJ
ijassa-372	69	19	words	word	NOUN
ijassa-372	69	20	by	by	ADP
ijassa-372	69	21	hard	hard	ADJ
ijassa-372	69	22	assignment	assignment	NOUN
ijassa-372	69	23	.	.	PUNCT
ijassa-372	70	1	many	many	ADJ
ijassa-372	70	2	more	more	ADV
ijassa-372	70	3	effective	effective	ADJ
ijassa-372	70	4	dictionary	dictionary	ADJ
ijassa-372	70	5	construction	construction	NOUN
ijassa-372	70	6	method	method	NOUN
ijassa-372	70	7	such	such	ADJ
ijassa-372	70	8	as	as	ADP
ijassa-372	70	9	hierarchical	hierarchical	ADJ
ijassa-372	70	10	clustering	clustering	NOUN
ijassa-372	70	11	,	,	PUNCT
ijassa-372	70	12	and	and	CCONJ
ijassa-372	70	13	more	more	ADV
ijassa-372	70	14	advanced	advanced	ADJ
ijassa-372	70	15	coding	code	VERB
ijassa-372	70	16	schemes	scheme	NOUN
ijassa-372	70	17	,	,	PUNCT
ijassa-372	70	18	such	such	ADJ
ijassa-372	70	19	as	as	ADP
ijassa-372	70	20	soft	soft	ADJ
ijassa-372	70	21	assignment	assignment	NOUN
ijassa-372	70	22	[	[	X
ijassa-372	70	23	17	17	NUM
ijassa-372	70	24	]	]	PUNCT
ijassa-372	70	25	,	,	PUNCT
ijassa-372	70	26	sparse	sparse	ADJ
ijassa-372	70	27	coding	code	VERB
ijassa-372	70	28	[	[	X
ijassa-372	70	29	5	5	NUM
ijassa-372	70	30	]	]	PUNCT
ijassa-372	70	31	,	,	PUNCT
ijassa-372	70	32	fisher	fisher	PROPN
ijassa-372	70	33	coding	code	VERB
ijassa-372	71	1	[	[	X
ijassa-372	71	2	18	18	NUM
ijassa-372	71	3	]	]	SYM
ijassa-372	71	4	et.al	et.al	PROPN
ijassa-372	71	5	.	.	PUNCT
ijassa-372	71	6	,	,	PUNCT
ijassa-372	71	7	all	all	PRON
ijassa-372	71	8	can	can	AUX
ijassa-372	71	9	be	be	AUX
ijassa-372	71	10	utilized	utilize	VERB
ijassa-372	71	11	in	in	ADP
ijassa-372	71	12	our	our	PRON
ijassa-372	71	13	method	method	NOUN
ijassa-372	71	14	.	.	PUNCT
ijassa-372	72	1	however	however	ADV
ijassa-372	72	2	,	,	PUNCT
ijassa-372	72	3	they	they	PRON
ijassa-372	72	4	are	be	AUX
ijassa-372	72	5	not	not	PART
ijassa-372	72	6	key	key	ADJ
ijassa-372	72	7	for	for	ADP
ijassa-372	72	8	evaluating	evaluate	VERB
ijassa-372	72	9	our	our	PRON
ijassa-372	72	10	method	method	NOUN
ijassa-372	72	11	,	,	PUNCT
ijassa-372	72	12	so	so	SCONJ
ijassa-372	72	13	we	we	PRON
ijassa-372	72	14	adopt	adopt	VERB
ijassa-372	72	15	the	the	DET
ijassa-372	72	16	easiest	easy	ADJ
ijassa-372	72	17	clustering	clustering	ADJ
ijassa-372	72	18	algorithm	algorithm	NOUN
ijassa-372	72	19	and	and	CCONJ
ijassa-372	72	20	coding	coding	NOUN
ijassa-372	72	21	method	method	NOUN
ijassa-372	72	22	.	.	PUNCT
ijassa-372	73	1	consequently	consequently	ADV
ijassa-372	73	2	,	,	PUNCT
ijassa-372	73	3	the	the	DET
ijassa-372	73	4	signature	signature	NOUN
ijassa-372	73	5	for	for	ADP
ijassa-372	73	6	representing	represent	VERB
ijassa-372	73	7	the	the	DET
ijassa-372	73	8	content	content	NOUN
ijassa-372	73	9	of	of	ADP
ijassa-372	73	10	an	an	DET
ijassa-372	73	11	image	image	NOUN
ijassa-372	73	12	is	be	AUX
ijassa-372	73	13	composed	compose	VERB
ijassa-372	73	14	of	of	ADP
ijassa-372	73	15	n	n	PRON
ijassa-372	73	16	patch	patch	NOUN
ijassa-372	73	17	features	feature	NOUN
ijassa-372	73	18	.	.	PUNCT
ijassa-372	74	1	the	the	DET
ijassa-372	74	2	i	i	PROPN
ijassa-372	74	3	-	-	PUNCT
ijassa-372	74	4	th	th	VERB
ijassa-372	74	5	patch	patch	NOUN
ijassa-372	74	6	feature	feature	NOUN
ijassa-372	74	7	is	be	AUX
ijassa-372	74	8	represented	represent	VERB
ijassa-372	74	9	as	as	ADP
ijassa-372	74	10	fi	fi	NOUN
ijassa-372	74	11	=	=	NOUN
ijassa-372	74	12	{	{	PUNCT
ijassa-372	74	13	ai	ai	NOUN
ijassa-372	74	14	,	,	PUNCT
ijassa-372	74	15	pi	pi	NOUN
ijassa-372	74	16	}	}	PUNCT
ijassa-372	74	17	,	,	PUNCT
ijassa-372	74	18	where	where	SCONJ
ijassa-372	74	19	ai	ai	ADV
ijassa-372	74	20	codes	code	NOUN
ijassa-372	74	21	the	the	DET
ijassa-372	74	22	appearance	appearance	NOUN
ijassa-372	74	23	feature	feature	NOUN
ijassa-372	74	24	in	in	ADP
ijassa-372	74	25	the	the	DET
ijassa-372	74	26	patch	patch	NOUN
ijassa-372	74	27	,	,	PUNCT
ijassa-372	74	28	whose	whose	DET
ijassa-372	74	29	dimension	dimension	NOUN
ijassa-372	74	30	is	be	AUX
ijassa-372	74	31	the	the	DET
ijassa-372	74	32	same	same	ADJ
ijassa-372	74	33	as	as	ADP
ijassa-372	74	34	the	the	DET
ijassa-372	74	35	size	size	NOUN
ijassa-372	74	36	of	of	ADP
ijassa-372	74	37	visual	visual	ADJ
ijassa-372	74	38	dictionary	dictionary	PROPN
ijassa-372	74	39	.	.	PUNCT
ijassa-372	75	1	pi	pi	NOUN
ijassa-372	75	2	=	=	SYM
ijassa-372	75	3	(	(	PUNCT
ijassa-372	75	4	xi	xi	PROPN
ijassa-372	75	5	,	,	PUNCT
ijassa-372	75	6	yi	yi	PROPN
ijassa-372	75	7	)	)	PUNCT
ijassa-372	75	8	is	be	AUX
ijassa-372	75	9	composed	compose	VERB
ijassa-372	75	10	of	of	ADP
ijassa-372	75	11	the	the	DET
ijassa-372	75	12	horizontal	horizontal	ADJ
ijassa-372	75	13	and	and	CCONJ
ijassa-372	75	14	vertical	vertical	ADJ
ijassa-372	75	15	spatial	spatial	ADJ
ijassa-372	75	16	positions	position	NOUN
ijassa-372	75	17	of	of	ADP
ijassa-372	75	18	the	the	DET
ijassa-372	75	19	patch	patch	NOUN
ijassa-372	75	20	in	in	ADP
ijassa-372	75	21	the	the	DET
ijassa-372	75	22	2d	2d	NUM
ijassa-372	75	23	plane	plane	NOUN
ijassa-372	75	24	space	space	NOUN
ijassa-372	75	25	.	.	PUNCT
ijassa-372	76	1	note	note	VERB
ijassa-372	76	2	that	that	SCONJ
ijassa-372	76	3	xi	xi	PROPN
ijassa-372	76	4	and	and	CCONJ
ijassa-372	76	5	yi	yi	PROPN
ijassa-372	76	6	both	both	PRON
ijassa-372	76	7	are	be	AUX
ijassa-372	76	8	normalized	normalize	VERB
ijassa-372	76	9	to	to	ADP
ijassa-372	76	10	[	[	X
ijassa-372	76	11	0,1	0,1	NUM
ijassa-372	76	12	]	]	PUNCT
ijassa-372	76	13	for	for	ADP
ijassa-372	76	14	each	each	DET
ijassa-372	76	15	patch	patch	NOUN
ijassa-372	76	16	.	.	PUNCT
ijassa-372	77	1	3.2	3.2	NUM
ijassa-372	77	2	distance	distance	NOUN
ijassa-372	77	3	between	between	ADP
ijassa-372	77	4	images	image	NOUN
ijassa-372	77	5	the	the	DET
ijassa-372	77	6	method	method	NOUN
ijassa-372	77	7	to	to	PART
ijassa-372	77	8	measure	measure	VERB
ijassa-372	77	9	the	the	DET
ijassa-372	77	10	distance	distance	NOUN
ijassa-372	77	11	between	between	ADP
ijassa-372	77	12	images	image	NOUN
ijassa-372	77	13	is	be	AUX
ijassa-372	77	14	critical	critical	ADJ
ijassa-372	77	15	for	for	ADP
ijassa-372	77	16	retrieval	retrieval	NOUN
ijassa-372	77	17	precision	precision	NOUN
ijassa-372	77	18	.	.	PUNCT
ijassa-372	78	1	in	in	ADP
ijassa-372	78	2	this	this	DET
ijassa-372	78	3	work	work	NOUN
ijassa-372	78	4	,	,	PUNCT
ijassa-372	78	5	a	a	DET
ijassa-372	78	6	flexible	flexible	ADJ
ijassa-372	78	7	patch	patch	NOUN
ijassa-372	78	8	-	-	PUNCT
ijassa-372	78	9	based	base	VERB
ijassa-372	78	10	distance	distance	NOUN
ijassa-372	78	11	measure	measure	NOUN
ijassa-372	78	12	method	method	NOUN
ijassa-372	78	13	is	be	AUX
ijassa-372	78	14	proposed	propose	VERB
ijassa-372	78	15	.	.	PUNCT
ijassa-372	79	1	given	give	VERB
ijassa-372	79	2	the	the	DET
ijassa-372	79	3	i	i	PROPN
ijassa-372	79	4	-	-	PUNCT
ijassa-372	79	5	th	th	X
ijassa-372	79	6	patch	patch	NOUN
ijassa-372	79	7	of	of	ADP
ijassa-372	79	8	the	the	DET
ijassa-372	79	9	first	first	ADJ
ijassa-372	79	10	image	image	NOUN
ijassa-372	79	11	u	u	NOUN
ijassa-372	79	12	and	and	CCONJ
ijassa-372	79	13	the	the	DET
ijassa-372	79	14	j	j	PROPN
ijassa-372	79	15	-	-	PUNCT
ijassa-372	79	16	th	th	X
ijassa-372	79	17	patch	patch	NOUN
ijassa-372	79	18	of	of	ADP
ijassa-372	79	19	the	the	DET
ijassa-372	79	20	second	second	ADJ
ijassa-372	79	21	image	image	NOUN
ijassa-372	79	22	246	246	NUM
ijassa-372	79	23	yan	yan	PROPN
ijassa-372	79	24	yu	yu	PROPN
ijassa-372	79	25	:	:	PUNCT
ijassa-372	79	26	image	image	NOUN
ijassa-372	79	27	retrieval	retrieval	NOUN
ijassa-372	79	28	based	base	VERB
ijassa-372	79	29	on	on	ADP
ijassa-372	79	30	spatial	spatial	ADJ
ijassa-372	79	31	organization	organization	NOUN
ijassa-372	79	32	of	of	ADP
ijassa-372	79	33	patches	patch	NOUN
ijassa-372	79	34	v	v	ADP
ijassa-372	79	35	,	,	PUNCT
ijassa-372	79	36	the	the	DET
ijassa-372	79	37	distance	distance	NOUN
ijassa-372	79	38	between	between	ADP
ijassa-372	79	39	patches	patch	NOUN
ijassa-372	79	40	is	be	AUX
ijassa-372	79	41	given	give	VERB
ijassa-372	79	42	by	by	ADP
ijassa-372	79	43	d(fu	d(fu	PROPN
ijassa-372	79	44	i	i	PRON
ijassa-372	79	45	,	,	PUNCT
ijassa-372	79	46	f	f	PROPN
ijassa-372	79	47	v	v	X
ijassa-372	79	48	j	j	PROPN
ijassa-372	79	49	)	)	PUNCT
ijassa-372	80	1	=	=	PUNCT
ijassa-372	80	2	wa	wa	X
ijassa-372	80	3	·	·	PUNCT
ijassa-372	80	4	d(au	d(au	PROPN
ijassa-372	81	1	i	i	PRON
ijassa-372	81	2	,	,	PUNCT
ijassa-372	81	3	a	a	DET
ijassa-372	81	4	v	v	X
ijassa-372	81	5	j	j	NOUN
ijassa-372	81	6	)	)	PUNCT
ijassa-372	81	7	δa	δa	PROPN
ijassa-372	81	8	+	+	NUM
ijassa-372	81	9	wp	wp	NOUN
ijassa-372	81	10	·	·	PUNCT
ijassa-372	81	11	d(p	d(p	PROPN
ijassa-372	81	12	u	u	NOUN
ijassa-372	81	13	i	i	PRON
ijassa-372	81	14	,	,	PUNCT
ijassa-372	81	15	p	p	PROPN
ijassa-372	81	16	v	v	X
ijassa-372	81	17	j	j	PROPN
ijassa-372	81	18	)	)	PUNCT
ijassa-372	81	19	δp	δp	ADV
ijassa-372	81	20	(	(	PUNCT
ijassa-372	81	21	2	2	NUM
ijassa-372	81	22	)	)	PUNCT
ijassa-372	81	23	where	where	SCONJ
ijassa-372	81	24	d(au	d(au	PROPN
ijassa-372	81	25	i	i	PRON
ijassa-372	81	26	,	,	PUNCT
ijassa-372	81	27	a	a	DET
ijassa-372	81	28	v	v	NOUN
ijassa-372	81	29	j	j	NOUN
ijassa-372	81	30	)	)	PUNCT
ijassa-372	81	31	is	be	AUX
ijassa-372	81	32	the	the	DET
ijassa-372	81	33	distance	distance	NOUN
ijassa-372	81	34	between	between	ADP
ijassa-372	81	35	appearance	appearance	NOUN
ijassa-372	81	36	features	feature	NOUN
ijassa-372	81	37	which	which	PRON
ijassa-372	81	38	can	can	AUX
ijassa-372	81	39	be	be	AUX
ijassa-372	81	40	computed	compute	VERB
ijassa-372	81	41	by	by	ADP
ijassa-372	81	42	chi	chi	ADJ
ijassa-372	81	43	-	-	PUNCT
ijassa-372	81	44	square	square	ADJ
ijassa-372	81	45	distance	distance	NOUN
ijassa-372	81	46	as	as	ADP
ijassa-372	81	47	(	(	PUNCT
ijassa-372	81	48	3	3	NUM
ijassa-372	81	49	)	)	PUNCT
ijassa-372	81	50	d(au	d(au	NOUN
ijassa-372	82	1	i	i	PRON
ijassa-372	82	2	,	,	PUNCT
ijassa-372	82	3	a	a	DET
ijassa-372	82	4	v	v	NOUN
ijassa-372	82	5	j	j	NOUN
ijassa-372	82	6	)	)	PUNCT
ijassa-372	82	7	=	=	PUNCT
ijassa-372	83	1	d∑	d∑	PROPN
ijassa-372	83	2	k=1	k=1	X
ijassa-372	84	1	(	(	PUNCT
ijassa-372	84	2	au	au	INTJ
ijassa-372	84	3	i	i	PROPN
ijassa-372	84	4	(	(	PUNCT
ijassa-372	84	5	k)−av	k)−av	PROPN
ijassa-372	84	6	j	j	PROPN
ijassa-372	84	7	(	(	PUNCT
ijassa-372	84	8	k	k	NOUN
ijassa-372	84	9	)	)	PUNCT
ijassa-372	84	10	)	)	PUNCT
ijassa-372	84	11	2	2	NUM
ijassa-372	84	12	au	au	VERB
ijassa-372	84	13	i	i	PRON
ijassa-372	84	14	(	(	PUNCT
ijassa-372	84	15	k	k	NOUN
ijassa-372	84	16	)	)	PUNCT
ijassa-372	85	1	+	+	PROPN
ijassa-372	85	2	av	av	PROPN
ijassa-372	85	3	j	j	PROPN
ijassa-372	85	4	(	(	PUNCT
ijassa-372	85	5	k	k	NOUN
ijassa-372	85	6	)	)	PUNCT
ijassa-372	85	7	(	(	PUNCT
ijassa-372	85	8	3	3	X
ijassa-372	85	9	)	)	PUNCT
ijassa-372	85	10	where	where	SCONJ
ijassa-372	85	11	d	d	NOUN
ijassa-372	85	12	is	be	AUX
ijassa-372	85	13	the	the	DET
ijassa-372	85	14	length	length	NOUN
ijassa-372	85	15	of	of	ADP
ijassa-372	85	16	the	the	DET
ijassa-372	85	17	appearance	appearance	NOUN
ijassa-372	85	18	feature	feature	NOUN
ijassa-372	85	19	vector	vector	NOUN
ijassa-372	85	20	,	,	PUNCT
ijassa-372	85	21	au	au	ADP
ijassa-372	85	22	i	i	PRON
ijassa-372	85	23	(	(	PUNCT
ijassa-372	85	24	k	k	NOUN
ijassa-372	85	25	)	)	PUNCT
ijassa-372	85	26	represent	represent	VERB
ijassa-372	85	27	the	the	DET
ijassa-372	85	28	k	k	PROPN
ijassa-372	85	29	-	-	PUNCT
ijassa-372	85	30	th	th	VERB
ijassa-372	85	31	component	component	NOUN
ijassa-372	85	32	value	value	NOUN
ijassa-372	85	33	of	of	ADP
ijassa-372	85	34	the	the	DET
ijassa-372	85	35	vector	vector	NOUN
ijassa-372	85	36	.	.	PUNCT
ijassa-372	86	1	and	and	CCONJ
ijassa-372	86	2	d(p	d(p	PROPN
ijassa-372	86	3	u	u	NOUN
ijassa-372	86	4	i	i	PRON
ijassa-372	86	5	,	,	PUNCT
ijassa-372	86	6	p	p	PROPN
ijassa-372	86	7	v	v	X
ijassa-372	86	8	j	j	PROPN
ijassa-372	86	9	)	)	PUNCT
ijassa-372	86	10	is	be	AUX
ijassa-372	86	11	the	the	DET
ijassa-372	86	12	distance	distance	NOUN
ijassa-372	86	13	between	between	ADP
ijassa-372	86	14	the	the	DET
ijassa-372	86	15	spatial	spatial	ADJ
ijassa-372	86	16	positions	position	NOUN
ijassa-372	86	17	of	of	ADP
ijassa-372	86	18	the	the	DET
ijassa-372	86	19	two	two	NUM
ijassa-372	86	20	patches	patch	NOUN
ijassa-372	86	21	,	,	PUNCT
ijassa-372	86	22	then	then	ADV
ijassa-372	86	23	we	we	PRON
ijassa-372	86	24	have	have	VERB
ijassa-372	86	25	d(p	d(p	PROPN
ijassa-372	86	26	u	u	NOUN
ijassa-372	86	27	i	i	PRON
ijassa-372	86	28	,	,	PUNCT
ijassa-372	86	29	p	p	PROPN
ijassa-372	86	30	v	v	X
ijassa-372	86	31	j	j	PROPN
ijassa-372	86	32	)	)	PUNCT
ijassa-372	87	1	=	=	PUNCT
ijassa-372	88	1	√	√	NUM
ijassa-372	88	2	(	(	PUNCT
ijassa-372	88	3	xui	xui	NOUN
ijassa-372	88	4	−	−	PROPN
ijassa-372	89	1	xvj	xvj	NOUN
ijassa-372	89	2	)	)	PUNCT
ijassa-372	89	3	2	2	NUM
ijassa-372	90	1	+	+	CCONJ
ijassa-372	90	2	(	(	PUNCT
ijassa-372	90	3	yui	yui	PROPN
ijassa-372	90	4	−	−	PROPN
ijassa-372	90	5	yvj	yvj	NOUN
ijassa-372	90	6	)	)	PUNCT
ijassa-372	90	7	(	(	PUNCT
ijassa-372	90	8	4	4	X
ijassa-372	90	9	)	)	PUNCT
ijassa-372	90	10	parameters	parameter	NOUN
ijassa-372	90	11	wa	wa	ADJ
ijassa-372	90	12	and	and	CCONJ
ijassa-372	90	13	wp	wp	PROPN
ijassa-372	90	14	are	be	AUX
ijassa-372	90	15	weight	weight	NOUN
ijassa-372	90	16	coefficients	coefficient	NOUN
ijassa-372	90	17	,	,	PUNCT
ijassa-372	90	18	which	which	PRON
ijassa-372	90	19	balance	balance	VERB
ijassa-372	90	20	appearance	appearance	NOUN
ijassa-372	90	21	and	and	CCONJ
ijassa-372	90	22	spatial	spatial	ADJ
ijassa-372	90	23	contributions	contribution	NOUN
ijassa-372	90	24	respectively	respectively	ADV
ijassa-372	90	25	and	and	CCONJ
ijassa-372	90	26	the	the	DET
ijassa-372	90	27	sum	sum	NOUN
ijassa-372	90	28	of	of	ADP
ijassa-372	90	29	them	they	PRON
ijassa-372	90	30	always	always	ADV
ijassa-372	90	31	is	be	AUX
ijassa-372	90	32	1	1	NUM
ijassa-372	90	33	.	.	PUNCT
ijassa-372	91	1	the	the	DET
ijassa-372	91	2	proportion	proportion	NOUN
ijassa-372	91	3	of	of	ADP
ijassa-372	91	4	them	they	PRON
ijassa-372	91	5	has	have	VERB
ijassa-372	91	6	a	a	DET
ijassa-372	91	7	great	great	ADJ
ijassa-372	91	8	influence	influence	NOUN
ijassa-372	91	9	on	on	ADP
ijassa-372	91	10	the	the	DET
ijassa-372	91	11	performance	performance	NOUN
ijassa-372	91	12	of	of	ADP
ijassa-372	91	13	the	the	DET
ijassa-372	91	14	method	method	NOUN
ijassa-372	91	15	,	,	PUNCT
ijassa-372	91	16	which	which	PRON
ijassa-372	91	17	will	will	AUX
ijassa-372	91	18	be	be	AUX
ijassa-372	91	19	discussed	discuss	VERB
ijassa-372	91	20	in	in	ADP
ijassa-372	91	21	the	the	DET
ijassa-372	91	22	next	next	ADJ
ijassa-372	91	23	section	section	NOUN
ijassa-372	91	24	.	.	PUNCT
ijassa-372	92	1	parameters	parameter	NOUN
ijassa-372	92	2	δa	δa	PROPN
ijassa-372	93	1	and	and	CCONJ
ijassa-372	93	2	δp	δp	PROPN
ijassa-372	93	3	are	be	AUX
ijassa-372	93	4	chosen	choose	VERB
ijassa-372	93	5	according	accord	VERB
ijassa-372	93	6	to	to	ADP
ijassa-372	93	7	the	the	DET
ijassa-372	93	8	appearance	appearance	NOUN
ijassa-372	93	9	and	and	CCONJ
ijassa-372	93	10	spatial	spatial	ADJ
ijassa-372	93	11	dynamics	dynamic	NOUN
ijassa-372	93	12	.	.	PUNCT
ijassa-372	94	1	typical	typical	ADJ
ijassa-372	94	2	values	value	NOUN
ijassa-372	94	3	used	use	VERB
ijassa-372	94	4	in	in	ADP
ijassa-372	94	5	this	this	DET
ijassa-372	94	6	paper	paper	NOUN
ijassa-372	94	7	are	be	AUX
ijassa-372	94	8	δa	δa	PRON
ijassa-372	94	9	=	=	SYM
ijassa-372	94	10	2	2	NUM
ijassa-372	94	11	and	and	CCONJ
ijassa-372	94	12	δp	δp	ADV
ijassa-372	94	13	=	=	NOUN
ijassa-372	94	14	√	√	PROPN
ijassa-372	94	15	2	2	NUM
ijassa-372	94	16	.	.	PUNCT
ijassa-372	95	1	the	the	DET
ijassa-372	95	2	distance	distance	NOUN
ijassa-372	95	3	d(fu	d(fu	PROPN
ijassa-372	95	4	i	i	PRON
ijassa-372	95	5	,	,	PUNCT
ijassa-372	95	6	f	f	PROPN
ijassa-372	95	7	v	v	X
ijassa-372	95	8	j	j	PROPN
ijassa-372	95	9	)	)	PUNCT
ijassa-372	95	10	between	between	ADP
ijassa-372	95	11	two	two	NUM
ijassa-372	95	12	patches	patch	NOUN
ijassa-372	95	13	can	can	AUX
ijassa-372	95	14	be	be	AUX
ijassa-372	95	15	seen	see	VERB
ijassa-372	95	16	as	as	ADP
ijassa-372	95	17	matching	match	VERB
ijassa-372	95	18	cost	cost	NOUN
ijassa-372	95	19	,	,	PUNCT
ijassa-372	95	20	which	which	PRON
ijassa-372	95	21	is	be	AUX
ijassa-372	95	22	denoted	denote	VERB
ijassa-372	95	23	as	as	ADP
ijassa-372	95	24	cij	cij	PROPN
ijassa-372	95	25	.	.	PUNCT
ijassa-372	96	1	given	give	VERB
ijassa-372	96	2	the	the	DET
ijassa-372	96	3	set	set	NOUN
ijassa-372	96	4	of	of	ADP
ijassa-372	96	5	costs	cost	NOUN
ijassa-372	96	6	between	between	ADP
ijassa-372	96	7	all	all	DET
ijassa-372	96	8	pairs	pair	NOUN
ijassa-372	96	9	of	of	ADP
ijassa-372	96	10	patches	patch	NOUN
ijassa-372	96	11	in	in	ADP
ijassa-372	96	12	the	the	DET
ijassa-372	96	13	first	first	ADJ
ijassa-372	96	14	image	image	NOUN
ijassa-372	96	15	and	and	CCONJ
ijassa-372	96	16	in	in	ADP
ijassa-372	96	17	the	the	DET
ijassa-372	96	18	second	second	ADJ
ijassa-372	96	19	image	image	NOUN
ijassa-372	96	20	,	,	PUNCT
ijassa-372	96	21	the	the	DET
ijassa-372	96	22	minimum	minimum	NOUN
ijassa-372	96	23	of	of	ADP
ijassa-372	96	24	the	the	DET
ijassa-372	96	25	total	total	ADJ
ijassa-372	96	26	matching	matching	NOUN
ijassa-372	96	27	cost	cost	NOUN
ijassa-372	96	28	can	can	AUX
ijassa-372	96	29	be	be	AUX
ijassa-372	96	30	seen	see	VERB
ijassa-372	96	31	the	the	DET
ijassa-372	96	32	distance	distance	NOUN
ijassa-372	96	33	between	between	ADP
ijassa-372	96	34	two	two	NUM
ijassa-372	96	35	images	image	NOUN
ijassa-372	96	36	,	,	PUNCT
ijassa-372	96	37	subject	subject	ADJ
ijassa-372	96	38	to	to	ADP
ijassa-372	96	39	the	the	DET
ijassa-372	96	40	constraint	constraint	NOUN
ijassa-372	96	41	that	that	PRON
ijassa-372	96	42	the	the	DET
ijassa-372	96	43	matching	matching	NOUN
ijassa-372	96	44	be	be	AUX
ijassa-372	96	45	one	one	NUM
ijassa-372	96	46	-	-	PUNCT
ijassa-372	96	47	to	to	ADP
ijassa-372	96	48	one	one	NUM
ijassa-372	96	49	.	.	PUNCT
ijassa-372	97	1	then	then	ADV
ijassa-372	97	2	the	the	DET
ijassa-372	97	3	distance	distance	NOUN
ijassa-372	97	4	between	between	ADP
ijassa-372	97	5	image	image	NOUN
ijassa-372	97	6	u	u	NOUN
ijassa-372	97	7	and	and	CCONJ
ijassa-372	97	8	image	image	NOUN
ijassa-372	97	9	v	v	NOUN
ijassa-372	97	10	is	be	AUX
ijassa-372	97	11	given	give	VERB
ijassa-372	97	12	by	by	ADP
ijassa-372	97	13	d(u	d(u	PROPN
ijassa-372	97	14	,	,	PUNCT
ijassa-372	97	15	v	v	NOUN
ijassa-372	97	16	)	)	PUNCT
ijassa-372	97	17	=	=	SYM
ijassa-372	98	1	min	min	PROPN
ijassa-372	99	1	π	π	PROPN
ijassa-372	99	2	n∑	n∑	PROPN
ijassa-372	99	3	i=1	i=1	PROPN
ijassa-372	99	4	d(fu	d(fu	PROPN
ijassa-372	100	1	i	i	PRON
ijassa-372	100	2	,	,	PUNCT
ijassa-372	100	3	f	f	PROPN
ijassa-372	100	4	v	v	ADP
ijassa-372	100	5	π(i	π(i	NOUN
ijassa-372	100	6	)	)	PUNCT
ijassa-372	100	7	)	)	PUNCT
ijassa-372	100	8	(	(	PUNCT
ijassa-372	100	9	6	6	X
ijassa-372	100	10	)	)	PUNCT
ijassa-372	100	11	where	where	SCONJ
ijassa-372	100	12	π	π	PROPN
ijassa-372	100	13	is	be	AUX
ijassa-372	100	14	a	a	DET
ijassa-372	100	15	permutation	permutation	NOUN
ijassa-372	100	16	of	of	ADP
ijassa-372	100	17	{	{	PUNCT
ijassa-372	100	18	1	1	NUM
ijassa-372	100	19	,	,	PUNCT
ijassa-372	100	20	...	...	PUNCT
ijassa-372	100	21	,	,	PUNCT
ijassa-372	100	22	n	n	CCONJ
ijassa-372	100	23	}	}	PUNCT
ijassa-372	100	24	to	to	PART
ijassa-372	100	25	guarantee	guarantee	VERB
ijassa-372	100	26	the	the	DET
ijassa-372	100	27	constraint	constraint	NOUN
ijassa-372	100	28	.	.	PUNCT
ijassa-372	101	1	the	the	DET
ijassa-372	101	2	input	input	NOUN
ijassa-372	101	3	to	to	ADP
ijassa-372	101	4	the	the	DET
ijassa-372	101	5	assignment	assignment	NOUN
ijassa-372	101	6	problem	problem	NOUN
ijassa-372	101	7	is	be	AUX
ijassa-372	101	8	a	a	DET
ijassa-372	101	9	square	square	ADJ
ijassa-372	101	10	cost	cost	NOUN
ijassa-372	101	11	matrix	matrix	NOUN
ijassa-372	101	12	with	with	ADP
ijassa-372	101	13	entries	entry	NOUN
ijassa-372	101	14	cij	cij	PROPN
ijassa-372	101	15	.	.	PUNCT
ijassa-372	102	1	the	the	DET
ijassa-372	102	2	result	result	NOUN
ijassa-372	102	3	is	be	AUX
ijassa-372	102	4	a	a	DET
ijassa-372	102	5	permutation	permutation	NOUN
ijassa-372	102	6	π	π	NOUN
ijassa-372	102	7	such	such	ADJ
ijassa-372	102	8	that	that	SCONJ
ijassa-372	102	9	total	total	ADJ
ijassa-372	102	10	matching	matching	NOUN
ijassa-372	102	11	cost	cost	NOUN
ijassa-372	102	12	is	be	AUX
ijassa-372	102	13	minimized	minimize	VERB
ijassa-372	102	14	.	.	PUNCT
ijassa-372	103	1	this	this	PRON
ijassa-372	103	2	is	be	AUX
ijassa-372	103	3	an	an	DET
ijassa-372	103	4	instance	instance	NOUN
ijassa-372	103	5	of	of	ADP
ijassa-372	103	6	the	the	DET
ijassa-372	103	7	square	square	ADJ
ijassa-372	103	8	assignment	assignment	NOUN
ijassa-372	103	9	problem	problem	NOUN
ijassa-372	103	10	,	,	PUNCT
ijassa-372	103	11	which	which	PRON
ijassa-372	103	12	can	can	AUX
ijassa-372	103	13	be	be	AUX
ijassa-372	103	14	solved	solve	VERB
ijassa-372	103	15	using	use	VERB
ijassa-372	103	16	the	the	DET
ijassa-372	103	17	hungarian	hungarian	ADJ
ijassa-372	103	18	algorithm	algorithm	NOUN
ijassa-372	103	19	[	[	X
ijassa-372	103	20	19	19	NUM
ijassa-372	103	21	]	]	PUNCT
ijassa-372	103	22	.	.	PUNCT
ijassa-372	104	1	given	give	VERB
ijassa-372	104	2	a	a	DET
ijassa-372	104	3	cost	cost	NOUN
ijassa-372	104	4	matrix	matrix	NOUN
ijassa-372	104	5	to	to	PART
ijassa-372	104	6	find	find	VERB
ijassa-372	104	7	an	an	DET
ijassa-372	104	8	optimal	optimal	ADJ
ijassa-372	104	9	assignment	assignment	NOUN
ijassa-372	104	10	using	use	VERB
ijassa-372	104	11	the	the	DET
ijassa-372	104	12	hungarian	hungarian	ADJ
ijassa-372	104	13	algorithm	algorithm	NOUN
ijassa-372	104	14	is	be	AUX
ijassa-372	104	15	presented	present	VERB
ijassa-372	104	16	as	as	ADP
ijassa-372	104	17	follow	follow	NOUN
ijassa-372	104	18	:	:	PUNCT
ijassa-372	104	19	1	1	X
ijassa-372	104	20	)	)	PUNCT
ijassa-372	104	21	subtract	subtract	VERB
ijassa-372	104	22	the	the	DET
ijassa-372	104	23	smallest	small	ADJ
ijassa-372	104	24	entry	entry	NOUN
ijassa-372	104	25	in	in	ADP
ijassa-372	104	26	each	each	DET
ijassa-372	104	27	row	row	NOUN
ijassa-372	104	28	from	from	ADP
ijassa-372	104	29	all	all	DET
ijassa-372	104	30	the	the	DET
ijassa-372	104	31	entries	entry	NOUN
ijassa-372	104	32	of	of	ADP
ijassa-372	104	33	its	its	PRON
ijassa-372	104	34	row	row	NOUN
ijassa-372	104	35	.	.	PUNCT
ijassa-372	105	1	2	2	X
ijassa-372	105	2	)	)	PUNCT
ijassa-372	105	3	subtract	subtract	VERB
ijassa-372	105	4	the	the	DET
ijassa-372	105	5	smallest	small	ADJ
ijassa-372	105	6	entry	entry	NOUN
ijassa-372	105	7	in	in	ADP
ijassa-372	105	8	each	each	DET
ijassa-372	105	9	column	column	NOUN
ijassa-372	105	10	from	from	ADP
ijassa-372	105	11	all	all	DET
ijassa-372	105	12	the	the	DET
ijassa-372	105	13	entries	entry	NOUN
ijassa-372	105	14	of	of	ADP
ijassa-372	105	15	its	its	PRON
ijassa-372	105	16	column	column	NOUN
ijassa-372	105	17	.	.	PUNCT
ijassa-372	106	1	3	3	X
ijassa-372	106	2	)	)	PUNCT
ijassa-372	106	3	draw	draw	VERB
ijassa-372	106	4	lines	line	NOUN
ijassa-372	106	5	through	through	ADP
ijassa-372	106	6	appropriate	appropriate	ADJ
ijassa-372	106	7	rows	row	NOUN
ijassa-372	106	8	and	and	CCONJ
ijassa-372	106	9	columns	column	NOUN
ijassa-372	106	10	so	so	SCONJ
ijassa-372	106	11	that	that	SCONJ
ijassa-372	106	12	all	all	DET
ijassa-372	106	13	the	the	DET
ijassa-372	106	14	zero	zero	NUM
ijassa-372	106	15	entries	entry	NOUN
ijassa-372	106	16	of	of	ADP
ijassa-372	106	17	the	the	DET
ijassa-372	106	18	cost	cost	NOUN
ijassa-372	106	19	matrix	matrix	NOUN
ijassa-372	106	20	are	be	AUX
ijassa-372	106	21	covered	cover	VERB
ijassa-372	106	22	and	and	CCONJ
ijassa-372	106	23	the	the	DET
ijassa-372	106	24	minimum	minimum	ADJ
ijassa-372	106	25	number	number	NOUN
ijassa-372	106	26	of	of	ADP
ijassa-372	106	27	such	such	ADJ
ijassa-372	106	28	lines	line	NOUN
ijassa-372	106	29	is	be	AUX
ijassa-372	106	30	used	use	VERB
ijassa-372	106	31	.	.	PUNCT
ijassa-372	107	1	advances	advance	NOUN
ijassa-372	107	2	in	in	ADP
ijassa-372	107	3	systems	system	NOUN
ijassa-372	107	4	science	science	NOUN
ijassa-372	107	5	and	and	CCONJ
ijassa-372	107	6	application	application	NOUN
ijassa-372	107	7	(	(	PUNCT
ijassa-372	107	8	2015	2015	NUM
ijassa-372	107	9	)	)	PUNCT
ijassa-372	107	10	vol.15	vol.15	NOUN
ijassa-372	107	11	no.3	no.3	VERB
ijassa-372	107	12	247	247	NUM
ijassa-372	107	13	4	4	NUM
ijassa-372	107	14	)	)	PUNCT
ijassa-372	107	15	test	test	NOUN
ijassa-372	107	16	for	for	ADP
ijassa-372	107	17	optimality	optimality	NOUN
ijassa-372	107	18	:	:	PUNCT
ijassa-372	107	19	(	(	PUNCT
ijassa-372	107	20	i	i	NOUN
ijassa-372	107	21	)	)	PUNCT
ijassa-372	107	22	if	if	SCONJ
ijassa-372	107	23	the	the	DET
ijassa-372	107	24	minimum	minimum	ADJ
ijassa-372	107	25	number	number	NOUN
ijassa-372	107	26	of	of	ADP
ijassa-372	107	27	covering	cover	VERB
ijassa-372	107	28	lines	line	NOUN
ijassa-372	107	29	is	be	AUX
ijassa-372	107	30	n	n	PRON
ijassa-372	107	31	,	,	PUNCT
ijassa-372	107	32	an	an	DET
ijassa-372	107	33	optimal	optimal	ADJ
ijassa-372	107	34	assignment	assignment	NOUN
ijassa-372	107	35	of	of	ADP
ijassa-372	107	36	zeroes	zero	NOUN
ijassa-372	107	37	is	be	AUX
ijassa-372	107	38	possible	possible	ADJ
ijassa-372	107	39	and	and	CCONJ
ijassa-372	107	40	we	we	PRON
ijassa-372	107	41	are	be	AUX
ijassa-372	107	42	finished	finish	VERB
ijassa-372	107	43	.	.	PUNCT
ijassa-372	108	1	(	(	PUNCT
ijassa-372	108	2	ii	ii	NOUN
ijassa-372	108	3	)	)	PUNCT
ijassa-372	108	4	if	if	SCONJ
ijassa-372	108	5	the	the	DET
ijassa-372	108	6	minimum	minimum	ADJ
ijassa-372	108	7	number	number	NOUN
ijassa-372	108	8	of	of	ADP
ijassa-372	108	9	covering	cover	VERB
ijassa-372	108	10	lines	line	NOUN
ijassa-372	108	11	is	be	AUX
ijassa-372	108	12	less	less	ADJ
ijassa-372	108	13	than	than	ADP
ijassa-372	108	14	n	n	CCONJ
ijassa-372	108	15	,	,	PUNCT
ijassa-372	108	16	an	an	DET
ijassa-372	108	17	optimal	optimal	ADJ
ijassa-372	108	18	assignment	assignment	NOUN
ijassa-372	108	19	of	of	ADP
ijassa-372	108	20	zeroes	zero	NOUN
ijassa-372	108	21	is	be	AUX
ijassa-372	108	22	not	not	PART
ijassa-372	108	23	yet	yet	ADV
ijassa-372	108	24	possible.in	possible.in	X
ijassa-372	108	25	that	that	DET
ijassa-372	108	26	case	case	NOUN
ijassa-372	108	27	,	,	PUNCT
ijassa-372	108	28	proceed	proceed	VERB
ijassa-372	108	29	to	to	ADP
ijassa-372	108	30	5	5	NUM
ijassa-372	108	31	)	)	PUNCT
ijassa-372	108	32	.	.	PUNCT
ijassa-372	109	1	5	5	X
ijassa-372	109	2	)	)	PUNCT
ijassa-372	109	3	determine	determine	VERB
ijassa-372	109	4	the	the	DET
ijassa-372	109	5	smallest	small	ADJ
ijassa-372	109	6	entry	entry	NOUN
ijassa-372	109	7	not	not	PART
ijassa-372	109	8	covered	cover	VERB
ijassa-372	109	9	by	by	ADP
ijassa-372	109	10	any	any	DET
ijassa-372	109	11	line	line	NOUN
ijassa-372	109	12	.	.	PUNCT
ijassa-372	110	1	subtract	subtract	VERB
ijassa-372	110	2	this	this	DET
ijassa-372	110	3	entry	entry	NOUN
ijassa-372	110	4	from	from	ADP
ijassa-372	110	5	each	each	DET
ijassa-372	110	6	uncovered	uncover	VERB
ijassa-372	110	7	row	row	NOUN
ijassa-372	110	8	,	,	PUNCT
ijassa-372	110	9	and	and	CCONJ
ijassa-372	110	10	then	then	ADV
ijassa-372	110	11	add	add	VERB
ijassa-372	110	12	it	it	PRON
ijassa-372	110	13	to	to	ADP
ijassa-372	110	14	each	each	DET
ijassa-372	110	15	covered	cover	VERB
ijassa-372	110	16	column	column	NOUN
ijassa-372	110	17	.	.	PUNCT
ijassa-372	111	1	return	return	VERB
ijassa-372	111	2	to	to	ADP
ijassa-372	111	3	3	3	NUM
ijassa-372	111	4	)	)	PUNCT
ijassa-372	111	5	.	.	PUNCT
ijassa-372	112	1	compared	compare	VERB
ijassa-372	112	2	with	with	ADP
ijassa-372	112	3	the	the	DET
ijassa-372	112	4	traditional	traditional	ADJ
ijassa-372	112	5	patch	patch	NOUN
ijassa-372	112	6	based	base	VERB
ijassa-372	112	7	distance	distance	NOUN
ijassa-372	112	8	measure	measure	NOUN
ijassa-372	112	9	methods	method	NOUN
ijassa-372	112	10	,	,	PUNCT
ijassa-372	112	11	our	our	PRON
ijassa-372	112	12	method	method	NOUN
ijassa-372	112	13	can	can	AUX
ijassa-372	112	14	deal	deal	VERB
ijassa-372	112	15	with	with	ADP
ijassa-372	112	16	the	the	DET
ijassa-372	112	17	position	position	NOUN
ijassa-372	112	18	change	change	NOUN
ijassa-372	112	19	of	of	ADP
ijassa-372	112	20	the	the	DET
ijassa-372	112	21	foreground	foreground	NOUN
ijassa-372	112	22	objects	object	NOUN
ijassa-372	112	23	in	in	ADP
ijassa-372	112	24	the	the	DET
ijassa-372	112	25	image	image	NOUN
ijassa-372	112	26	automatically	automatically	ADV
ijassa-372	112	27	.	.	PUNCT
ijassa-372	113	1	we	we	PRON
ijassa-372	113	2	should	should	AUX
ijassa-372	113	3	note	note	VERB
ijassa-372	113	4	that	that	SCONJ
ijassa-372	113	5	the	the	DET
ijassa-372	113	6	distance	distance	NOUN
ijassa-372	113	7	that	that	PRON
ijassa-372	113	8	we	we	PRON
ijassa-372	113	9	proposed	propose	VERB
ijassa-372	113	10	is	be	AUX
ijassa-372	113	11	not	not	PART
ijassa-372	113	12	a	a	DET
ijassa-372	113	13	rigid	rigid	ADJ
ijassa-372	113	14	mathematical	mathematical	ADJ
ijassa-372	113	15	distance	distance	NOUN
ijassa-372	113	16	.	.	PUNCT
ijassa-372	114	1	however	however	ADV
ijassa-372	114	2	,	,	PUNCT
ijassa-372	114	3	it	it	PRON
ijassa-372	114	4	dose	dose	VERB
ijassa-372	114	5	not	not	PART
ijassa-372	114	6	matter	matter	NOUN
ijassa-372	114	7	because	because	SCONJ
ijassa-372	114	8	the	the	DET
ijassa-372	114	9	distance	distance	NOUN
ijassa-372	114	10	can	can	AUX
ijassa-372	114	11	be	be	AUX
ijassa-372	114	12	used	use	VERB
ijassa-372	114	13	to	to	PART
ijassa-372	114	14	represent	represent	VERB
ijassa-372	114	15	the	the	DET
ijassa-372	114	16	divergence	divergence	NOUN
ijassa-372	114	17	between	between	ADP
ijassa-372	114	18	images	image	NOUN
ijassa-372	114	19	and	and	CCONJ
ijassa-372	114	20	it	it	PRON
ijassa-372	114	21	can	can	AUX
ijassa-372	114	22	be	be	AUX
ijassa-372	114	23	utilized	utilize	VERB
ijassa-372	114	24	in	in	ADP
ijassa-372	114	25	ranking	rank	VERB
ijassa-372	114	26	retrieval	retrieval	NOUN
ijassa-372	114	27	results	result	NOUN
ijassa-372	114	28	.	.	PUNCT
ijassa-372	115	1	4	4	NUM
ijassa-372	115	2	experiments	experiment	NOUN
ijassa-372	115	3	in	in	ADP
ijassa-372	115	4	this	this	DET
ijassa-372	115	5	section	section	NOUN
ijassa-372	115	6	,	,	PUNCT
ijassa-372	115	7	we	we	PRON
ijassa-372	115	8	evaluate	evaluate	VERB
ijassa-372	115	9	the	the	DET
ijassa-372	115	10	performance	performance	NOUN
ijassa-372	115	11	of	of	ADP
ijassa-372	115	12	our	our	PRON
ijassa-372	115	13	proposed	propose	VERB
ijassa-372	115	14	method	method	NOUN
ijassa-372	115	15	in	in	ADP
ijassa-372	115	16	image	image	NOUN
ijassa-372	115	17	retrieval	retrieval	NOUN
ijassa-372	115	18	scenario	scenario	NOUN
ijassa-372	115	19	.	.	PUNCT
ijassa-372	116	1	4.1	4.1	NUM
ijassa-372	116	2	experimental	experimental	ADJ
ijassa-372	116	3	setups	setup	NOUN
ijassa-372	116	4	experiments	experiment	NOUN
ijassa-372	116	5	are	be	AUX
ijassa-372	116	6	performed	perform	VERB
ijassa-372	116	7	on	on	ADP
ijassa-372	116	8	a	a	DET
ijassa-372	116	9	personal	personal	ADJ
ijassa-372	116	10	laptop	laptop	NOUN
ijassa-372	116	11	in	in	ADP
ijassa-372	116	12	windows	window	NOUN
ijassa-372	116	13	8.1	8.1	NUM
ijassa-372	116	14	,	,	PUNCT
ijassa-372	116	15	with	with	ADP
ijassa-372	116	16	quad	quad	PROPN
ijassa-372	116	17	intel	intel	PROPN
ijassa-372	116	18	core	core	NOUN
ijassa-372	116	19	i7	i7	NOUN
ijassa-372	116	20	-	-	PUNCT
ijassa-372	116	21	4500u	4500u	NUM
ijassa-372	116	22	running	run	VERB
ijassa-372	116	23	at	at	ADP
ijassa-372	116	24	1.8ghz	1.8ghz	NUM
ijassa-372	116	25	.	.	PUNCT
ijassa-372	117	1	most	most	ADJ
ijassa-372	117	2	of	of	ADP
ijassa-372	117	3	our	our	PRON
ijassa-372	117	4	programs	program	NOUN
ijassa-372	117	5	are	be	AUX
ijassa-372	117	6	implemented	implement	VERB
ijassa-372	117	7	in	in	ADP
ijassa-372	117	8	matlab	matlab	PROPN
ijassa-372	117	9	except	except	SCONJ
ijassa-372	117	10	dense	dense	ADJ
ijassa-372	117	11	sift	sift	ADJ
ijassa-372	117	12	and	and	CCONJ
ijassa-372	117	13	hungarian	hungarian	ADJ
ijassa-372	117	14	algorithm	algorithm	NOUN
ijassa-372	117	15	for	for	ADP
ijassa-372	117	16	which	which	PRON
ijassa-372	117	17	we	we	PRON
ijassa-372	117	18	use	use	VERB
ijassa-372	117	19	open	open	ADJ
ijassa-372	117	20	source	source	NOUN
ijassa-372	117	21	code	code	NOUN
ijassa-372	117	22	in	in	ADP
ijassa-372	117	23	c.	c.	PROPN
ijassa-372	117	24	our	our	PRON
ijassa-372	117	25	dataset	dataset	NOUN
ijassa-372	117	26	is	be	AUX
ijassa-372	117	27	composed	compose	VERB
ijassa-372	117	28	of	of	ADP
ijassa-372	117	29	fifteen	fifteen	NUM
ijassa-372	117	30	scene	scene	NOUN
ijassa-372	117	31	categories	category	NOUN
ijassa-372	117	32	,	,	PUNCT
ijassa-372	117	33	including	include	VERB
ijassa-372	117	34	bedroom	bedroom	NOUN
ijassa-372	117	35	,	,	PUNCT
ijassa-372	117	36	kitchen	kitchen	NOUN
ijassa-372	117	37	,	,	PUNCT
ijassa-372	117	38	livingroom	livingroom	NOUN
ijassa-372	117	39	,	,	PUNCT
ijassa-372	117	40	store	store	NOUN
ijassa-372	117	41	,	,	PUNCT
ijassa-372	117	42	coast	coast	NOUN
ijassa-372	117	43	,	,	PUNCT
ijassa-372	117	44	forest	forest	NOUN
ijassa-372	117	45	,	,	PUNCT
ijassa-372	117	46	et.al	et.al	PROPN
ijassa-372	117	47	..	..	PUNCT
ijassa-372	117	48	each	each	DET
ijassa-372	117	49	category	category	NOUN
ijassa-372	117	50	contains	contain	VERB
ijassa-372	117	51	200	200	NUM
ijassa-372	117	52	to	to	PART
ijassa-372	117	53	400	400	NUM
ijassa-372	117	54	images	image	NOUN
ijassa-372	117	55	,	,	PUNCT
ijassa-372	117	56	and	and	CCONJ
ijassa-372	117	57	average	average	ADJ
ijassa-372	117	58	image	image	NOUN
ijassa-372	117	59	size	size	NOUN
ijassa-372	117	60	is	be	AUX
ijassa-372	117	61	300×250	300×250	NUM
ijassa-372	117	62	pixels	pixel	NOUN
ijassa-372	117	63	.	.	PUNCT
ijassa-372	118	1	the	the	DET
ijassa-372	118	2	dataset	dataset	NOUN
ijassa-372	118	3	is	be	AUX
ijassa-372	118	4	popular	popular	ADJ
ijassa-372	118	5	in	in	ADP
ijassa-372	118	6	image	image	NOUN
ijassa-372	118	7	classification	classification	NOUN
ijassa-372	118	8	and	and	CCONJ
ijassa-372	118	9	retrieval	retrieval	NOUN
ijassa-372	118	10	scenario	scenario	NOUN
ijassa-372	118	11	.	.	PUNCT
ijassa-372	119	1	fig.2	fig.2	PROPN
ijassa-372	119	2	show	show	VERB
ijassa-372	119	3	examples	example	NOUN
ijassa-372	119	4	of	of	ADP
ijassa-372	119	5	the	the	DET
ijassa-372	119	6	dataset	dataset	NOUN
ijassa-372	119	7	.	.	PUNCT
ijassa-372	120	1	we	we	PRON
ijassa-372	120	2	use	use	VERB
ijassa-372	120	3	100	100	NUM
ijassa-372	120	4	images	image	NOUN
ijassa-372	120	5	per	per	ADP
ijassa-372	120	6	class	class	NOUN
ijassa-372	120	7	randomly	randomly	ADV
ijassa-372	120	8	selected	select	VERB
ijassa-372	120	9	for	for	ADP
ijassa-372	120	10	generating	generate	VERB
ijassa-372	120	11	visual	visual	ADJ
ijassa-372	120	12	words	word	NOUN
ijassa-372	120	13	dictionary	dictionary	ADJ
ijassa-372	120	14	and	and	CCONJ
ijassa-372	120	15	the	the	DET
ijassa-372	120	16	rest	rest	NOUN
ijassa-372	120	17	for	for	ADP
ijassa-372	120	18	testing	testing	NOUN
ijassa-372	120	19	.	.	PUNCT
ijassa-372	121	1	we	we	PRON
ijassa-372	121	2	use	use	VERB
ijassa-372	121	3	dictionary	dictionary	ADJ
ijassa-372	121	4	size	size	NOUN
ijassa-372	121	5	200	200	NUM
ijassa-372	121	6	and	and	CCONJ
ijassa-372	121	7	hard	hard	ADJ
ijassa-372	121	8	assignment	assignment	NOUN
ijassa-372	121	9	for	for	ADP
ijassa-372	121	10	testing	test	VERB
ijassa-372	121	11	our	our	PRON
ijassa-372	121	12	method	method	NOUN
ijassa-372	121	13	.	.	PUNCT
ijassa-372	122	1	in	in	ADP
ijassa-372	122	2	practical	practical	ADJ
ijassa-372	122	3	application	application	NOUN
ijassa-372	122	4	,	,	PUNCT
ijassa-372	122	5	the	the	DET
ijassa-372	122	6	soft	soft	ADJ
ijassa-372	122	7	assignment	assignment	NOUN
ijassa-372	122	8	can	can	AUX
ijassa-372	122	9	be	be	AUX
ijassa-372	122	10	used	use	VERB
ijassa-372	122	11	in	in	ADP
ijassa-372	122	12	our	our	PRON
ijassa-372	122	13	framework	framework	NOUN
ijassa-372	122	14	for	for	ADP
ijassa-372	122	15	improving	improve	VERB
ijassa-372	122	16	retrieval	retrieval	NOUN
ijassa-372	122	17	performance	performance	NOUN
ijassa-372	122	18	.	.	PUNCT
ijassa-372	123	1	the	the	DET
ijassa-372	123	2	experimental	experimental	ADJ
ijassa-372	123	3	results	result	NOUN
ijassa-372	123	4	are	be	AUX
ijassa-372	123	5	presented	present	VERB
ijassa-372	123	6	in	in	ADP
ijassa-372	123	7	terms	term	NOUN
ijassa-372	123	8	of	of	ADP
ijassa-372	123	9	mean	mean	ADJ
ijassa-372	123	10	average	average	ADJ
ijassa-372	123	11	precision	precision	NOUN
ijassa-372	123	12	(	(	PUNCT
ijassa-372	123	13	map	map	NOUN
ijassa-372	123	14	)	)	PUNCT
ijassa-372	123	15	and	and	CCONJ
ijassa-372	123	16	precision	precision	NOUN
ijassa-372	123	17	for	for	ADP
ijassa-372	123	18	the	the	DET
ijassa-372	123	19	top	top	ADJ
ijassa-372	123	20	n	n	NOUN
ijassa-372	123	21	retrieved	retrieve	VERB
ijassa-372	123	22	images	image	NOUN
ijassa-372	123	23	(	(	PUNCT
ijassa-372	123	24	p@n	p@n	NOUN
ijassa-372	123	25	)	)	PUNCT
ijassa-372	123	26	.	.	PUNCT
ijassa-372	124	1	although	although	SCONJ
ijassa-372	124	2	map	map	NOUN
ijassa-372	124	3	is	be	AUX
ijassa-372	124	4	a	a	DET
ijassa-372	124	5	very	very	ADV
ijassa-372	124	6	popular	popular	ADJ
ijassa-372	124	7	measure	measure	NOUN
ijassa-372	124	8	to	to	PART
ijassa-372	124	9	assess	assess	VERB
ijassa-372	124	10	the	the	DET
ijassa-372	124	11	effectiveness	effectiveness	NOUN
ijassa-372	124	12	of	of	ADP
ijassa-372	124	13	cbir	cbir	NOUN
ijassa-372	124	14	methods	method	NOUN
ijassa-372	124	15	,	,	PUNCT
ijassa-372	124	16	it	it	PRON
ijassa-372	124	17	dose	dose	AUX
ijassa-372	124	18	not	not	PART
ijassa-372	124	19	reflect	reflect	VERB
ijassa-372	124	20	the	the	DET
ijassa-372	124	21	ranking	ranking	ADJ
ijassa-372	124	22	quality	quality	NOUN
ijassa-372	124	23	in	in	ADP
ijassa-372	124	24	the	the	DET
ijassa-372	124	25	first	first	ADJ
ijassa-372	124	26	positions	position	NOUN
ijassa-372	124	27	.	.	PUNCT
ijassa-372	125	1	the	the	DET
ijassa-372	125	2	practical	practical	ADJ
ijassa-372	125	3	web	web	NOUN
ijassa-372	125	4	users	user	NOUN
ijassa-372	125	5	usually	usually	ADV
ijassa-372	125	6	pay	pay	VERB
ijassa-372	125	7	more	more	ADJ
ijassa-372	125	8	attentions	attention	NOUN
ijassa-372	125	9	to	to	ADP
ijassa-372	125	10	the	the	DET
ijassa-372	125	11	good	good	ADJ
ijassa-372	125	12	set	set	NOUN
ijassa-372	125	13	of	of	ADP
ijassa-372	125	14	top	top	ADJ
ijassa-372	125	15	10	10	NUM
ijassa-372	125	16	or	or	CCONJ
ijassa-372	125	17	20	20	NUM
ijassa-372	125	18	retrieved	retrieve	VERB
ijassa-372	125	19	images	image	NOUN
ijassa-372	125	20	,	,	PUNCT
ijassa-372	125	21	so	so	SCONJ
ijassa-372	125	22	we	we	PRON
ijassa-372	125	23	are	be	AUX
ijassa-372	125	24	more	more	ADV
ijassa-372	125	25	interested	interested	ADJ
ijassa-372	125	26	in	in	ADP
ijassa-372	125	27	good	good	ADJ
ijassa-372	125	28	p@n	p@n	NOUN
ijassa-372	125	29	values	value	NOUN
ijassa-372	125	30	than	than	SCONJ
ijassa-372	125	31	map	map	VERB
ijassa-372	125	32	values	value	NOUN
ijassa-372	125	33	.	.	PUNCT
ijassa-372	126	1	precision	precision	NOUN
ijassa-372	126	2	measures	measure	VERB
ijassa-372	126	3	the	the	DET
ijassa-372	126	4	retrieval	retrieval	NOUN
ijassa-372	126	5	accuracy	accuracy	NOUN
ijassa-372	126	6	.	.	PUNCT
ijassa-372	127	1	it	it	PRON
ijassa-372	127	2	is	be	AUX
ijassa-372	127	3	defined	define	VERB
ijassa-372	127	4	as	as	ADP
ijassa-372	127	5	the	the	DET
ijassa-372	127	6	ratio	ratio	NOUN
ijassa-372	127	7	between	between	ADP
ijassa-372	127	8	the	the	DET
ijassa-372	127	9	number	number	NOUN
ijassa-372	127	10	of	of	ADP
ijassa-372	127	11	relevant	relevant	ADJ
ijassa-372	127	12	images	image	NOUN
ijassa-372	127	13	retrieved	retrieve	VERB
ijassa-372	127	14	and	and	CCONJ
ijassa-372	127	15	the	the	DET
ijassa-372	127	16	total	total	ADJ
ijassa-372	127	17	number	number	NOUN
ijassa-372	127	18	of	of	ADP
ijassa-372	127	19	images	image	NOUN
ijassa-372	127	20	retrieved	retrieve	VERB
ijassa-372	127	21	.	.	PUNCT
ijassa-372	128	1	relevant	relevant	ADJ
ijassa-372	128	2	images	image	NOUN
ijassa-372	128	3	are	be	AUX
ijassa-372	128	4	in	in	ADP
ijassa-372	128	5	the	the	DET
ijassa-372	128	6	same	same	ADJ
ijassa-372	128	7	category	category	NOUN
ijassa-372	128	8	with	with	ADP
ijassa-372	128	9	the	the	DET
ijassa-372	128	10	query	query	NOUN
ijassa-372	128	11	example	example	NOUN
ijassa-372	128	12	.	.	PUNCT
ijassa-372	129	1	248	248	NUM
ijassa-372	129	2	yan	yan	PROPN
ijassa-372	129	3	yu	yu	PROPN
ijassa-372	129	4	:	:	PUNCT
ijassa-372	129	5	image	image	NOUN
ijassa-372	129	6	retrieval	retrieval	NOUN
ijassa-372	129	7	based	base	VERB
ijassa-372	129	8	on	on	ADP
ijassa-372	129	9	spatial	spatial	ADJ
ijassa-372	129	10	organization	organization	NOUN
ijassa-372	129	11	of	of	ADP
ijassa-372	129	12	patches	patch	NOUN
ijassa-372	129	13	(	(	PUNCT
ijassa-372	129	14	a)livingroom	a)livingroom	NOUN
ijassa-372	129	15	(	(	PUNCT
ijassa-372	129	16	b)kitchen	b)kitchen	PROPN
ijassa-372	129	17	(	(	PUNCT
ijassa-372	129	18	c)bedroom	c)bedroom	NOUN
ijassa-372	129	19	(	(	PUNCT
ijassa-372	129	20	d)mitcoast	d)mitcoast	PROPN
ijassa-372	129	21	(	(	PUNCT
ijassa-372	129	22	e)industrial	e)industrial	PROPN
ijassa-372	129	23	(	(	PUNCT
ijassa-372	129	24	f)mitmountain	f)mitmountain	VERB
ijassa-372	129	25	fig.2	fig.2	ADJ
ijassa-372	129	26	sample	sample	NOUN
ijassa-372	129	27	images	image	NOUN
ijassa-372	129	28	from	from	ADP
ijassa-372	129	29	the	the	DET
ijassa-372	129	30	dataset	dataset	NOUN
ijassa-372	129	31	,	,	PUNCT
ijassa-372	129	32	highlighting	highlight	VERB
ijassa-372	129	33	6	6	NUM
ijassa-372	129	34	categories	category	NOUN
ijassa-372	129	35	(	(	PUNCT
ijassa-372	129	36	livingroom	livingroom	NOUN
ijassa-372	129	37	,	,	PUNCT
ijassa-372	129	38	kitchen	kitchen	NOUN
ijassa-372	129	39	,	,	PUNCT
ijassa-372	129	40	bedroom	bedroom	NOUN
ijassa-372	129	41	,	,	PUNCT
ijassa-372	129	42	mitcoast	mitcoast	NOUN
ijassa-372	129	43	,	,	PUNCT
ijassa-372	129	44	industrial	industrial	ADJ
ijassa-372	129	45	and	and	CCONJ
ijassa-372	129	46	mitmountain	mitmountain	NOUN
ijassa-372	129	47	)	)	PUNCT
ijassa-372	129	48	.	.	PUNCT
ijassa-372	130	1	table	table	NOUN
ijassa-372	130	2	1	1	NUM
ijassa-372	130	3	comparison	comparison	NOUN
ijassa-372	130	4	between	between	ADP
ijassa-372	130	5	two	two	NUM
ijassa-372	130	6	kinds	kind	NOUN
ijassa-372	130	7	of	of	ADP
ijassa-372	130	8	partition	partition	NOUN
ijassa-372	130	9	patterns	pattern	NOUN
ijassa-372	130	10	based	base	VERB
ijassa-372	130	11	on	on	ADP
ijassa-372	130	12	10	10	NUM
ijassa-372	130	13	queries	query	NOUN
ijassa-372	130	14	.	.	PUNCT
ijassa-372	131	1	partition	partition	NOUN
ijassa-372	131	2	pattern	pattern	NOUN
ijassa-372	131	3	p@5(%	p@5(%	NOUN
ijassa-372	131	4	)	)	PUNCT
ijassa-372	131	5	p@10(%	p@10(%	ADJ
ijassa-372	131	6	)	)	PUNCT
ijassa-372	131	7	average	average	ADJ
ijassa-372	131	8	retrieval	retrieval	NOUN
ijassa-372	131	9	time(ms	time(ms	NOUN
ijassa-372	131	10	)	)	PUNCT
ijassa-372	131	11	4×	4×	NOUN
ijassa-372	131	12	4	4	NUM
ijassa-372	131	13	62.00	62.00	NUM
ijassa-372	131	14	63.00	63.00	NUM
ijassa-372	131	15	23.64	23.64	NUM
ijassa-372	131	16	8×	8×	NUM
ijassa-372	131	17	8	8	NUM
ijassa-372	131	18	78.00	78.00	NUM
ijassa-372	131	19	69.00	69.00	NUM
ijassa-372	131	20	185.77	185.77	NUM
ijassa-372	131	21	table	table	NOUN
ijassa-372	131	22	2	2	NUM
ijassa-372	131	23	comparison	comparison	NOUN
ijassa-372	131	24	between	between	ADP
ijassa-372	131	25	two	two	NUM
ijassa-372	131	26	kinds	kind	NOUN
ijassa-372	131	27	of	of	ADP
ijassa-372	131	28	weighting	weight	VERB
ijassa-372	131	29	ratios	ratio	NOUN
ijassa-372	131	30	based	base	VERB
ijassa-372	131	31	on	on	ADP
ijassa-372	131	32	100	100	NUM
ijassa-372	131	33	queries	query	NOUN
ijassa-372	131	34	.	.	PUNCT
ijassa-372	132	1	weighting	weight	VERB
ijassa-372	132	2	ratio	ratio	NOUN
ijassa-372	132	3	p@5(%	p@5(%	PROPN
ijassa-372	132	4	)	)	PUNCT
ijassa-372	132	5	p@10(%	p@10(%	ADJ
ijassa-372	132	6	)	)	PUNCT
ijassa-372	132	7	p@15(%	p@15(%	NOUN
ijassa-372	132	8	)	)	PUNCT
ijassa-372	132	9	p@20(%	p@20(%	NOUN
ijassa-372	132	10	)	)	PUNCT
ijassa-372	132	11	1:1	1:1	NUM
ijassa-372	133	1	67.00	67.00	NUM
ijassa-372	133	2	59.10	59.10	NUM
ijassa-372	133	3	55.47	55.47	NUM
ijassa-372	133	4	52.50	52.50	NUM
ijassa-372	133	5	2:1	2:1	NUM
ijassa-372	133	6	67.40	67.40	NUM
ijassa-372	133	7	59.90	59.90	NUM
ijassa-372	133	8	55.80	55.80	NUM
ijassa-372	133	9	53.65	53.65	NUM
ijassa-372	133	10	4:1	4:1	NUM
ijassa-372	133	11	68.60	68.60	NUM
ijassa-372	133	12	60.00	60.00	NUM
ijassa-372	133	13	57.00	57.00	NUM
ijassa-372	133	14	55.15	55.15	NUM
ijassa-372	133	15	8:1	8:1	NUM
ijassa-372	133	16	66.40	66.40	NUM
ijassa-372	133	17	59.80	59.80	NUM
ijassa-372	133	18	56.33	56.33	NUM
ijassa-372	133	19	54.60	54.60	NUM
ijassa-372	133	20	4.2	4.2	NUM
ijassa-372	133	21	influence	influence	NOUN
ijassa-372	133	22	about	about	ADP
ijassa-372	133	23	size	size	NOUN
ijassa-372	133	24	of	of	ADP
ijassa-372	133	25	patches	patch	NOUN
ijassa-372	133	26	in	in	ADP
ijassa-372	133	27	this	this	DET
ijassa-372	133	28	section	section	NOUN
ijassa-372	133	29	,	,	PUNCT
ijassa-372	133	30	the	the	DET
ijassa-372	133	31	impact	impact	NOUN
ijassa-372	133	32	of	of	ADP
ijassa-372	133	33	the	the	DET
ijassa-372	133	34	patch	patch	ADJ
ijassa-372	133	35	size	size	NOUN
ijassa-372	133	36	is	be	AUX
ijassa-372	133	37	illustrated	illustrate	VERB
ijassa-372	133	38	.	.	PUNCT
ijassa-372	134	1	we	we	PRON
ijassa-372	134	2	use	use	VERB
ijassa-372	134	3	n	n	NOUN
ijassa-372	134	4	to	to	PART
ijassa-372	134	5	present	present	VERB
ijassa-372	134	6	the	the	DET
ijassa-372	134	7	total	total	ADJ
ijassa-372	134	8	number	number	NOUN
ijassa-372	134	9	of	of	ADP
ijassa-372	134	10	patches	patch	NOUN
ijassa-372	134	11	in	in	ADP
ijassa-372	134	12	each	each	DET
ijassa-372	134	13	image	image	NOUN
ijassa-372	134	14	.	.	PUNCT
ijassa-372	135	1	we	we	PRON
ijassa-372	135	2	divide	divide	VERB
ijassa-372	135	3	each	each	DET
ijassa-372	135	4	images	image	NOUN
ijassa-372	135	5	into	into	ADP
ijassa-372	135	6	several	several	ADJ
ijassa-372	135	7	rows	row	NOUN
ijassa-372	135	8	and	and	CCONJ
ijassa-372	135	9	several	several	ADJ
ijassa-372	135	10	columns	column	NOUN
ijassa-372	135	11	,	,	PUNCT
ijassa-372	135	12	including	include	VERB
ijassa-372	135	13	two	two	NUM
ijassa-372	135	14	partition	partition	NOUN
ijassa-372	135	15	patterns	pattern	NOUN
ijassa-372	135	16	which	which	PRON
ijassa-372	135	17	are	be	AUX
ijassa-372	135	18	4	4	NUM
ijassa-372	135	19	×	×	NOUN
ijassa-372	135	20	4	4	NUM
ijassa-372	135	21	and	and	CCONJ
ijassa-372	135	22	8	8	NUM
ijassa-372	135	23	×	×	NOUN
ijassa-372	135	24	8	8	NUM
ijassa-372	135	25	in	in	ADP
ijassa-372	135	26	our	our	PRON
ijassa-372	135	27	experiments	experiment	NOUN
ijassa-372	135	28	.	.	PUNCT
ijassa-372	136	1	to	to	PART
ijassa-372	136	2	compare	compare	VERB
ijassa-372	136	3	the	the	DET
ijassa-372	136	4	two	two	NUM
ijassa-372	136	5	pattern	pattern	NOUN
ijassa-372	136	6	patterns	pattern	NOUN
ijassa-372	136	7	,	,	PUNCT
ijassa-372	136	8	10	10	NUM
ijassa-372	136	9	images	image	NOUN
ijassa-372	136	10	are	be	AUX
ijassa-372	136	11	randomly	randomly	ADV
ijassa-372	136	12	selected	select	VERB
ijassa-372	136	13	from	from	ADP
ijassa-372	136	14	the	the	DET
ijassa-372	136	15	dataset	dataset	NOUN
ijassa-372	136	16	and	and	CCONJ
ijassa-372	136	17	each	each	PRON
ijassa-372	136	18	of	of	ADP
ijassa-372	136	19	them	they	PRON
ijassa-372	136	20	is	be	AUX
ijassa-372	136	21	iteratively	iteratively	ADV
ijassa-372	136	22	selected	select	VERB
ijassa-372	136	23	as	as	ADP
ijassa-372	136	24	a	a	DET
ijassa-372	136	25	advances	advance	NOUN
ijassa-372	136	26	in	in	ADP
ijassa-372	136	27	systems	system	NOUN
ijassa-372	136	28	science	science	NOUN
ijassa-372	136	29	and	and	CCONJ
ijassa-372	136	30	application	application	NOUN
ijassa-372	136	31	(	(	PUNCT
ijassa-372	136	32	2015	2015	NUM
ijassa-372	136	33	)	)	PUNCT
ijassa-372	136	34	vol.15	vol.15	NOUN
ijassa-372	136	35	no.3	no.3	VERB
ijassa-372	136	36	249	249	NUM
ijassa-372	136	37	query	query	NOUN
ijassa-372	136	38	example	example	NOUN
ijassa-372	136	39	,	,	PUNCT
ijassa-372	136	40	so	so	CCONJ
ijassa-372	136	41	the	the	DET
ijassa-372	136	42	total	total	ADJ
ijassa-372	136	43	number	number	NOUN
ijassa-372	136	44	of	of	ADP
ijassa-372	136	45	queries	query	NOUN
ijassa-372	136	46	in	in	ADP
ijassa-372	136	47	each	each	DET
ijassa-372	136	48	pattern	pattern	NOUN
ijassa-372	136	49	is	be	AUX
ijassa-372	136	50	10	10	NUM
ijassa-372	136	51	.	.	PUNCT
ijassa-372	137	1	it	it	PRON
ijassa-372	137	2	is	be	AUX
ijassa-372	137	3	obvious	obvious	ADJ
ijassa-372	137	4	that	that	SCONJ
ijassa-372	137	5	the	the	DET
ijassa-372	137	6	size	size	NOUN
ijassa-372	137	7	of	of	ADP
ijassa-372	137	8	each	each	DET
ijassa-372	137	9	patch	patch	NOUN
ijassa-372	137	10	is	be	AUX
ijassa-372	137	11	decreasing	decrease	VERB
ijassa-372	137	12	when	when	SCONJ
ijassa-372	137	13	the	the	DET
ijassa-372	137	14	total	total	ADJ
ijassa-372	137	15	number	number	NOUN
ijassa-372	137	16	of	of	ADP
ijassa-372	137	17	the	the	DET
ijassa-372	137	18	patches	patch	NOUN
ijassa-372	137	19	is	be	AUX
ijassa-372	137	20	increasing	increase	VERB
ijassa-372	137	21	.	.	PUNCT
ijassa-372	138	1	the	the	DET
ijassa-372	138	2	details	detail	NOUN
ijassa-372	138	3	can	can	AUX
ijassa-372	138	4	be	be	AUX
ijassa-372	138	5	seen	see	VERB
ijassa-372	138	6	in	in	ADP
ijassa-372	138	7	table	table	NOUN
ijassa-372	138	8	1	1	NUM
ijassa-372	138	9	.	.	PUNCT
ijassa-372	139	1	it	it	PRON
ijassa-372	139	2	can	can	AUX
ijassa-372	139	3	be	be	AUX
ijassa-372	139	4	seen	see	VERB
ijassa-372	139	5	that	that	SCONJ
ijassa-372	139	6	increasing	increase	VERB
ijassa-372	139	7	the	the	DET
ijassa-372	139	8	total	total	ADJ
ijassa-372	139	9	number	number	NOUN
ijassa-372	139	10	of	of	ADP
ijassa-372	139	11	the	the	DET
ijassa-372	139	12	patches	patch	NOUN
ijassa-372	139	13	has	have	VERB
ijassa-372	139	14	a	a	DET
ijassa-372	139	15	improvent	improvent	NOUN
ijassa-372	139	16	on	on	ADP
ijassa-372	139	17	the	the	DET
ijassa-372	139	18	retrieval	retrieval	NOUN
ijassa-372	139	19	results	result	NOUN
ijassa-372	139	20	,	,	PUNCT
ijassa-372	139	21	however	however	ADV
ijassa-372	139	22	,	,	PUNCT
ijassa-372	139	23	the	the	DET
ijassa-372	139	24	computation	computation	NOUN
ijassa-372	139	25	become	become	VERB
ijassa-372	139	26	more	more	ADJ
ijassa-372	139	27	time	time	NOUN
ijassa-372	139	28	-	-	PUNCT
ijassa-372	139	29	consuming	consume	VERB
ijassa-372	139	30	.	.	PUNCT
ijassa-372	140	1	in	in	ADP
ijassa-372	140	2	consideration	consideration	NOUN
ijassa-372	140	3	of	of	ADP
ijassa-372	140	4	the	the	DET
ijassa-372	140	5	efficiency	efficiency	NOUN
ijassa-372	140	6	,	,	PUNCT
ijassa-372	140	7	in	in	ADP
ijassa-372	140	8	the	the	DET
ijassa-372	140	9	rest	rest	NOUN
ijassa-372	140	10	of	of	ADP
ijassa-372	140	11	the	the	DET
ijassa-372	140	12	paper	paper	NOUN
ijassa-372	140	13	patches	patch	NOUN
ijassa-372	140	14	of	of	ADP
ijassa-372	140	15	n=16	n=16	NOUN
ijassa-372	140	16	are	be	AUX
ijassa-372	140	17	always	always	ADV
ijassa-372	140	18	used	use	VERB
ijassa-372	140	19	.	.	PUNCT
ijassa-372	141	1	4.3	4.3	NUM
ijassa-372	141	2	appearance	appearance	NOUN
ijassa-372	141	3	-	-	PUNCT
ijassa-372	141	4	spatial	spatial	NOUN
ijassa-372	141	5	weighting	weight	VERB
ijassa-372	141	6	the	the	DET
ijassa-372	141	7	weighting	weighting	NOUN
ijassa-372	141	8	parameters	parameter	NOUN
ijassa-372	141	9	wa	wa	ADJ
ijassa-372	141	10	and	and	CCONJ
ijassa-372	141	11	wp	wp	PROPN
ijassa-372	141	12	balance	balance	NOUN
ijassa-372	141	13	appearance	appearance	NOUN
ijassa-372	141	14	feature	feature	NOUN
ijassa-372	141	15	and	and	CCONJ
ijassa-372	141	16	spatial	spatial	ADJ
ijassa-372	141	17	contributions	contribution	NOUN
ijassa-372	141	18	respectively	respectively	ADV
ijassa-372	141	19	.	.	PUNCT
ijassa-372	142	1	in	in	ADP
ijassa-372	142	2	this	this	DET
ijassa-372	142	3	section	section	NOUN
ijassa-372	142	4	,	,	PUNCT
ijassa-372	142	5	the	the	DET
ijassa-372	142	6	impact	impact	NOUN
ijassa-372	142	7	of	of	ADP
ijassa-372	142	8	ratio	ratio	NOUN
ijassa-372	142	9	of	of	ADP
ijassa-372	142	10	the	the	DET
ijassa-372	142	11	weighting	weighting	NOUN
ijassa-372	142	12	parameters	parameter	NOUN
ijassa-372	142	13	is	be	AUX
ijassa-372	142	14	illustrated	illustrate	VERB
ijassa-372	142	15	.	.	PUNCT
ijassa-372	143	1	to	to	PART
ijassa-372	143	2	compare	compare	VERB
ijassa-372	143	3	the	the	DET
ijassa-372	143	4	influence	influence	NOUN
ijassa-372	143	5	of	of	ADP
ijassa-372	143	6	weighting	weight	VERB
ijassa-372	143	7	parameters	parameter	NOUN
ijassa-372	143	8	,	,	PUNCT
ijassa-372	143	9	100	100	NUM
ijassa-372	143	10	images	image	NOUN
ijassa-372	143	11	are	be	AUX
ijassa-372	143	12	randomly	randomly	ADV
ijassa-372	143	13	selected	select	VERB
ijassa-372	143	14	from	from	ADP
ijassa-372	143	15	the	the	DET
ijassa-372	143	16	dataset	dataset	NOUN
ijassa-372	143	17	and	and	CCONJ
ijassa-372	143	18	each	each	PRON
ijassa-372	143	19	of	of	ADP
ijassa-372	143	20	them	they	PRON
ijassa-372	143	21	is	be	AUX
ijassa-372	143	22	iteratively	iteratively	ADV
ijassa-372	143	23	selected	select	VERB
ijassa-372	143	24	as	as	ADP
ijassa-372	143	25	a	a	DET
ijassa-372	143	26	query	query	NOUN
ijassa-372	143	27	example	example	NOUN
ijassa-372	143	28	,	,	PUNCT
ijassa-372	143	29	so	so	CCONJ
ijassa-372	143	30	the	the	DET
ijassa-372	143	31	total	total	ADJ
ijassa-372	143	32	number	number	NOUN
ijassa-372	143	33	of	of	ADP
ijassa-372	143	34	queries	query	NOUN
ijassa-372	143	35	in	in	ADP
ijassa-372	143	36	each	each	DET
ijassa-372	143	37	experiment	experiment	NOUN
ijassa-372	143	38	is	be	AUX
ijassa-372	143	39	100	100	NUM
ijassa-372	143	40	.	.	PUNCT
ijassa-372	144	1	if	if	SCONJ
ijassa-372	144	2	we	we	PRON
ijassa-372	144	3	wish	wish	VERB
ijassa-372	144	4	each	each	DET
ijassa-372	144	5	patch	patch	NOUN
ijassa-372	144	6	to	to	PART
ijassa-372	144	7	be	be	AUX
ijassa-372	144	8	matched	match	VERB
ijassa-372	144	9	with	with	ADP
ijassa-372	144	10	the	the	DET
ijassa-372	144	11	patch	patch	NOUN
ijassa-372	144	12	which	which	PRON
ijassa-372	144	13	is	be	AUX
ijassa-372	144	14	close	close	ADJ
ijassa-372	144	15	to	to	ADP
ijassa-372	144	16	it	it	PRON
ijassa-372	144	17	in	in	ADP
ijassa-372	144	18	spatial	spatial	ADJ
ijassa-372	144	19	position	position	NOUN
ijassa-372	144	20	,	,	PUNCT
ijassa-372	144	21	wp	wp	PROPN
ijassa-372	144	22	must	must	AUX
ijassa-372	144	23	be	be	AUX
ijassa-372	144	24	set	set	VERB
ijassa-372	144	25	at	at	ADP
ijassa-372	144	26	a	a	DET
ijassa-372	144	27	high	high	ADJ
ijassa-372	144	28	value	value	NOUN
ijassa-372	144	29	.	.	PUNCT
ijassa-372	145	1	otherwise	otherwise	ADV
ijassa-372	145	2	,	,	PUNCT
ijassa-372	145	3	if	if	SCONJ
ijassa-372	145	4	we	we	PRON
ijassa-372	145	5	wish	wish	VERB
ijassa-372	145	6	each	each	DET
ijassa-372	145	7	patch	patch	NOUN
ijassa-372	145	8	in	in	ADP
ijassa-372	145	9	the	the	DET
ijassa-372	145	10	first	first	ADJ
ijassa-372	145	11	image	image	NOUN
ijassa-372	145	12	can	can	AUX
ijassa-372	145	13	be	be	AUX
ijassa-372	145	14	matches	match	NOUN
ijassa-372	145	15	with	with	ADP
ijassa-372	145	16	any	any	DET
ijassa-372	145	17	patch	patch	NOUN
ijassa-372	145	18	in	in	ADP
ijassa-372	145	19	the	the	DET
ijassa-372	145	20	second	second	ADJ
ijassa-372	145	21	image	image	NOUN
ijassa-372	145	22	,	,	PUNCT
ijassa-372	145	23	wp	wp	PROPN
ijassa-372	145	24	must	must	AUX
ijassa-372	145	25	be	be	AUX
ijassa-372	145	26	set	set	VERB
ijassa-372	145	27	at	at	ADP
ijassa-372	145	28	a	a	DET
ijassa-372	145	29	low	low	ADJ
ijassa-372	145	30	value	value	NOUN
ijassa-372	145	31	.	.	PUNCT
ijassa-372	146	1	we	we	PRON
ijassa-372	146	2	test	test	VERB
ijassa-372	146	3	the	the	DET
ijassa-372	146	4	weighting	weighting	NOUN
ijassa-372	146	5	parameters	parameter	NOUN
ijassa-372	146	6	including	include	VERB
ijassa-372	146	7	four	four	NUM
ijassa-372	146	8	kinds	kind	NOUN
ijassa-372	146	9	of	of	ADP
ijassa-372	146	10	ratio	ratio	NOUN
ijassa-372	146	11	patterns	pattern	NOUN
ijassa-372	146	12	which	which	PRON
ijassa-372	146	13	are	be	AUX
ijassa-372	146	14	1:1	1:1	NUM
ijassa-372	146	15	,	,	PUNCT
ijassa-372	146	16	2:1	2:1	NUM
ijassa-372	146	17	,	,	PUNCT
ijassa-372	146	18	4:1	4:1	NUM
ijassa-372	146	19	and	and	CCONJ
ijassa-372	146	20	8:1	8:1	NUM
ijassa-372	146	21	.	.	PUNCT
ijassa-372	147	1	the	the	DET
ijassa-372	147	2	details	detail	NOUN
ijassa-372	147	3	can	can	AUX
ijassa-372	147	4	be	be	AUX
ijassa-372	147	5	seen	see	VERB
ijassa-372	147	6	in	in	ADP
ijassa-372	147	7	table2	table2	PROPN
ijassa-372	147	8	.	.	PUNCT
ijassa-372	148	1	it	it	PRON
ijassa-372	148	2	can	can	AUX
ijassa-372	148	3	be	be	AUX
ijassa-372	148	4	seen	see	VERB
ijassa-372	148	5	that	that	SCONJ
ijassa-372	148	6	increasing	increase	VERB
ijassa-372	148	7	the	the	DET
ijassa-372	148	8	proportion	proportion	NOUN
ijassa-372	148	9	of	of	ADP
ijassa-372	148	10	wa	wa	NOUN
ijassa-372	148	11	has	have	VERB
ijassa-372	148	12	a	a	DET
ijassa-372	148	13	slight	slight	ADJ
ijassa-372	148	14	improvement	improvement	NOUN
ijassa-372	148	15	on	on	ADP
ijassa-372	148	16	results	result	NOUN
ijassa-372	148	17	.	.	PUNCT
ijassa-372	149	1	but	but	CCONJ
ijassa-372	149	2	if	if	SCONJ
ijassa-372	149	3	the	the	DET
ijassa-372	149	4	ratio	ratio	NOUN
ijassa-372	149	5	keep	keep	VERB
ijassa-372	149	6	going	go	VERB
ijassa-372	149	7	up	up	ADP
ijassa-372	149	8	and	and	CCONJ
ijassa-372	149	9	reach	reach	VERB
ijassa-372	149	10	8:1	8:1	NUM
ijassa-372	149	11	,	,	PUNCT
ijassa-372	149	12	the	the	DET
ijassa-372	149	13	precision	precision	NOUN
ijassa-372	149	14	begin	begin	VERB
ijassa-372	149	15	to	to	PART
ijassa-372	149	16	decline	decline	VERB
ijassa-372	149	17	.	.	PUNCT
ijassa-372	150	1	in	in	ADP
ijassa-372	150	2	the	the	DET
ijassa-372	150	3	rest	rest	NOUN
ijassa-372	150	4	of	of	ADP
ijassa-372	150	5	the	the	DET
ijassa-372	150	6	paper	paper	NOUN
ijassa-372	150	7	the	the	DET
ijassa-372	150	8	ratio	ratio	NOUN
ijassa-372	150	9	4:1	4:1	NUM
ijassa-372	150	10	are	be	AUX
ijassa-372	150	11	always	always	ADV
ijassa-372	150	12	used	use	VERB
ijassa-372	150	13	.	.	PUNCT
ijassa-372	151	1	4.4	4.4	NUM
ijassa-372	151	2	comparison	comparison	NOUN
ijassa-372	151	3	with	with	ADP
ijassa-372	151	4	the	the	DET
ijassa-372	151	5	traditional	traditional	ADJ
ijassa-372	151	6	method	method	NOUN
ijassa-372	151	7	the	the	DET
ijassa-372	151	8	traditional	traditional	ADJ
ijassa-372	151	9	bow	bow	NOUN
ijassa-372	151	10	method	method	NOUN
ijassa-372	151	11	is	be	AUX
ijassa-372	151	12	also	also	ADV
ijassa-372	151	13	implemented	implement	VERB
ijassa-372	151	14	for	for	ADP
ijassa-372	151	15	comparison	comparison	NOUN
ijassa-372	151	16	.	.	PUNCT
ijassa-372	152	1	to	to	PART
ijassa-372	152	2	compare	compare	VERB
ijassa-372	152	3	our	our	PRON
ijassa-372	152	4	proposed	propose	VERB
ijassa-372	152	5	method	method	NOUN
ijassa-372	152	6	with	with	ADP
ijassa-372	152	7	it	it	PRON
ijassa-372	152	8	fairly	fairly	ADV
ijassa-372	152	9	,	,	PUNCT
ijassa-372	152	10	we	we	PRON
ijassa-372	152	11	iteratively	iteratively	ADV
ijassa-372	152	12	select	select	VERB
ijassa-372	152	13	each	each	DET
ijassa-372	152	14	image	image	NOUN
ijassa-372	152	15	in	in	ADP
ijassa-372	152	16	the	the	DET
ijassa-372	152	17	dataset	dataset	NOUN
ijassa-372	152	18	as	as	ADP
ijassa-372	152	19	query	query	NOUN
ijassa-372	152	20	example	example	NOUN
ijassa-372	152	21	,	,	PUNCT
ijassa-372	152	22	so	so	CCONJ
ijassa-372	152	23	the	the	DET
ijassa-372	152	24	total	total	ADJ
ijassa-372	152	25	number	number	NOUN
ijassa-372	152	26	of	of	ADP
ijassa-372	152	27	queries	query	NOUN
ijassa-372	152	28	is	be	AUX
ijassa-372	152	29	2985	2985	NUM
ijassa-372	152	30	in	in	ADP
ijassa-372	152	31	each	each	DET
ijassa-372	152	32	query	query	NOUN
ijassa-372	152	33	experiment	experiment	NOUN
ijassa-372	152	34	.	.	PUNCT
ijassa-372	153	1	in	in	ADP
ijassa-372	153	2	table	table	NOUN
ijassa-372	153	3	3	3	NUM
ijassa-372	153	4	,	,	PUNCT
ijassa-372	153	5	it	it	PRON
ijassa-372	153	6	can	can	AUX
ijassa-372	153	7	be	be	AUX
ijassa-372	153	8	observed	observe	VERB
ijassa-372	153	9	that	that	SCONJ
ijassa-372	153	10	the	the	DET
ijassa-372	153	11	precision	precision	NOUN
ijassa-372	153	12	decreases	decrease	VERB
ijassa-372	153	13	while	while	SCONJ
ijassa-372	153	14	the	the	DET
ijassa-372	153	15	number	number	NOUN
ijassa-372	153	16	of	of	ADP
ijassa-372	153	17	retrieved	retrieve	VERB
ijassa-372	153	18	images	image	NOUN
ijassa-372	153	19	increases	increase	NOUN
ijassa-372	153	20	.	.	PUNCT
ijassa-372	154	1	in	in	ADP
ijassa-372	154	2	our	our	PRON
ijassa-372	154	3	proposed	propose	VERB
ijassa-372	154	4	method	method	NOUN
ijassa-372	154	5	,	,	PUNCT
ijassa-372	154	6	the	the	DET
ijassa-372	154	7	p@n	p@n	NOUN
ijassa-372	154	8	is	be	AUX
ijassa-372	154	9	always	always	ADV
ijassa-372	154	10	better	well	ADJ
ijassa-372	154	11	than	than	ADP
ijassa-372	154	12	the	the	DET
ijassa-372	154	13	traditional	traditional	ADJ
ijassa-372	154	14	bow	bow	NOUN
ijassa-372	154	15	method	method	NOUN
ijassa-372	154	16	.	.	PUNCT
ijassa-372	155	1	therefore	therefore	ADV
ijassa-372	155	2	,	,	PUNCT
ijassa-372	155	3	our	our	PRON
ijassa-372	155	4	proposed	propose	VERB
ijassa-372	155	5	method	method	NOUN
ijassa-372	155	6	is	be	AUX
ijassa-372	155	7	more	more	ADV
ijassa-372	155	8	suitable	suitable	ADJ
ijassa-372	155	9	for	for	ADP
ijassa-372	155	10	image	image	NOUN
ijassa-372	155	11	retrieval	retrieval	NOUN
ijassa-372	155	12	.	.	PUNCT
ijassa-372	156	1	table	table	NOUN
ijassa-372	156	2	3	3	NUM
ijassa-372	156	3	comparison	comparison	NOUN
ijassa-372	156	4	with	with	ADP
ijassa-372	156	5	the	the	DET
ijassa-372	156	6	traditional	traditional	ADJ
ijassa-372	156	7	bow	bow	NOUN
ijassa-372	156	8	based	base	VERB
ijassa-372	156	9	on	on	ADP
ijassa-372	156	10	2985	2985	NUM
ijassa-372	156	11	queries	query	NOUN
ijassa-372	156	12	.	.	PUNCT
ijassa-372	157	1	methods	method	NOUN
ijassa-372	157	2	p@5(%	p@5(%	PROPN
ijassa-372	157	3	)	)	PUNCT
ijassa-372	157	4	p@10(%	p@10(%	ADJ
ijassa-372	157	5	)	)	PUNCT
ijassa-372	157	6	p@15(%	p@15(%	NOUN
ijassa-372	157	7	)	)	PUNCT
ijassa-372	157	8	p@20(%	p@20(%	NOUN
ijassa-372	157	9	)	)	PUNCT
ijassa-372	157	10	map(%	map(%	NOUN
ijassa-372	157	11	)	)	PUNCT
ijassa-372	157	12	traditional	traditional	ADJ
ijassa-372	157	13	bow	bow	NOUN
ijassa-372	157	14	57.45	57.45	NUM
ijassa-372	157	15	50.67	50.67	NUM
ijassa-372	157	16	47.92	47.92	NUM
ijassa-372	157	17	46.08	46.08	NUM
ijassa-372	157	18	29.28	29.28	NUM
ijassa-372	157	19	our	our	PRON
ijassa-372	157	20	method	method	NOUN
ijassa-372	157	21	66.11	66.11	NUM
ijassa-372	157	22	58.87	58.87	NUM
ijassa-372	157	23	55.30	55.30	NUM
ijassa-372	157	24	53.15	53.15	NUM
ijassa-372	157	25	32.54	32.54	NUM
ijassa-372	157	26	in	in	ADP
ijassa-372	157	27	the	the	DET
ijassa-372	157	28	end	end	NOUN
ijassa-372	157	29	,	,	PUNCT
ijassa-372	157	30	we	we	PRON
ijassa-372	157	31	give	give	VERB
ijassa-372	157	32	some	some	DET
ijassa-372	157	33	retrieval	retrieval	NOUN
ijassa-372	157	34	examples	example	NOUN
ijassa-372	157	35	of	of	ADP
ijassa-372	157	36	the	the	DET
ijassa-372	157	37	query	query	NOUN
ijassa-372	157	38	image	image	NOUN
ijassa-372	157	39	random	random	NOUN
ijassa-372	157	40	selected	select	VERB
ijassa-372	157	41	from	from	ADP
ijassa-372	157	42	the	the	DET
ijassa-372	157	43	dataset	dataset	NOUN
ijassa-372	157	44	via	via	ADP
ijassa-372	157	45	the	the	DET
ijassa-372	157	46	two	two	NUM
ijassa-372	157	47	kinds	kind	NOUN
ijassa-372	157	48	of	of	ADP
ijassa-372	157	49	methods	method	NOUN
ijassa-372	157	50	in	in	ADP
ijassa-372	157	51	fig	fig	NOUN
ijassa-372	157	52	.	.	PUNCT
ijassa-372	158	1	3	3	NUM
ijassa-372	158	2	,	,	PUNCT
ijassa-372	158	3	fig	fig	NOUN
ijassa-372	158	4	.	.	NOUN
ijassa-372	158	5	4	4	NUM
ijassa-372	158	6	and	and	CCONJ
ijassa-372	158	7	fig.5	fig.5	PROPN
ijassa-372	158	8	.	.	PUNCT
ijassa-372	159	1	the	the	DET
ijassa-372	159	2	query	query	NOUN
ijassa-372	159	3	images	image	NOUN
ijassa-372	159	4	come	come	VERB
ijassa-372	159	5	from	from	ADP
ijassa-372	159	6	“	"	PUNCT
ijassa-372	159	7	calsuburb	calsuburb	NOUN
ijassa-372	159	8	”	"	PUNCT
ijassa-372	159	9	,	,	PUNCT
ijassa-372	159	10	“	"	PUNCT
ijassa-372	159	11	mittallbuilding”and	mittallbuilding”and	CCONJ
ijassa-372	159	12	“	"	PUNCT
ijassa-372	159	13	mitmountain	mitmountain	NOUN
ijassa-372	159	14	”	"	PUNCT
ijassa-372	159	15	250	250	NUM
ijassa-372	159	16	yan	yan	PROPN
ijassa-372	159	17	yu	yu	PROPN
ijassa-372	159	18	:	:	PUNCT
ijassa-372	159	19	image	image	NOUN
ijassa-372	159	20	retrieval	retrieval	NOUN
ijassa-372	159	21	based	base	VERB
ijassa-372	159	22	on	on	ADP
ijassa-372	159	23	spatial	spatial	ADJ
ijassa-372	159	24	organization	organization	NOUN
ijassa-372	159	25	of	of	ADP
ijassa-372	159	26	patches	patch	NOUN
ijassa-372	159	27	category	category	NOUN
ijassa-372	159	28	respectively	respectively	ADV
ijassa-372	159	29	.	.	PUNCT
ijassa-372	160	1	in	in	ADP
ijassa-372	160	2	each	each	DET
ijassa-372	160	3	fig	fig	NOUN
ijassa-372	160	4	.	.	PUNCT
ijassa-372	161	1	,	,	PUNCT
ijassa-372	161	2	the	the	DET
ijassa-372	161	3	query	query	NOUN
ijassa-372	161	4	image	image	NOUN
ijassa-372	161	5	lies	lie	VERB
ijassa-372	161	6	on	on	ADP
ijassa-372	161	7	the	the	DET
ijassa-372	161	8	top	top	ADJ
ijassa-372	161	9	left	leave	VERB
ijassa-372	161	10	corner	corner	NOUN
ijassa-372	161	11	.	.	PUNCT
ijassa-372	162	1	retrieved	retrieve	VERB
ijassa-372	162	2	top	top	ADJ
ijassa-372	162	3	12	12	NUM
ijassa-372	162	4	images	image	NOUN
ijassa-372	162	5	are	be	AUX
ijassa-372	162	6	arranged	arrange	VERB
ijassa-372	162	7	from	from	ADP
ijassa-372	162	8	left	leave	VERB
ijassa-372	162	9	to	to	ADP
ijassa-372	162	10	right	right	NOUN
ijassa-372	162	11	and	and	CCONJ
ijassa-372	162	12	top	top	VERB
ijassa-372	162	13	to	to	ADP
ijassa-372	162	14	bottom	bottom	NOUN
ijassa-372	162	15	according	accord	VERB
ijassa-372	162	16	to	to	ADP
ijassa-372	162	17	the	the	DET
ijassa-372	162	18	increasing	increase	VERB
ijassa-372	162	19	distances	distance	NOUN
ijassa-372	162	20	between	between	ADP
ijassa-372	162	21	the	the	DET
ijassa-372	162	22	retrieved	retrieve	VERB
ijassa-372	162	23	images	image	NOUN
ijassa-372	162	24	and	and	CCONJ
ijassa-372	162	25	the	the	DET
ijassa-372	162	26	query	query	NOUN
ijassa-372	162	27	example	example	NOUN
ijassa-372	162	28	.	.	PUNCT
ijassa-372	163	1	it	it	PRON
ijassa-372	163	2	can	can	AUX
ijassa-372	163	3	be	be	AUX
ijassa-372	163	4	seen	see	VERB
ijassa-372	163	5	that	that	SCONJ
ijassa-372	163	6	the	the	DET
ijassa-372	163	7	vast	vast	ADJ
ijassa-372	163	8	majority	majority	NOUN
ijassa-372	163	9	of	of	ADP
ijassa-372	163	10	the	the	DET
ijassa-372	163	11	retrieved	retrieve	VERB
ijassa-372	163	12	images	image	NOUN
ijassa-372	163	13	are	be	AUX
ijassa-372	163	14	in	in	ADP
ijassa-372	163	15	the	the	DET
ijassa-372	163	16	same	same	ADJ
ijassa-372	163	17	category	category	NOUN
ijassa-372	163	18	with	with	ADP
ijassa-372	163	19	the	the	DET
ijassa-372	163	20	query	query	NOUN
ijassa-372	163	21	image	image	NOUN
ijassa-372	163	22	in	in	ADP
ijassa-372	163	23	our	our	PRON
ijassa-372	163	24	proposed	propose	VERB
ijassa-372	163	25	method	method	NOUN
ijassa-372	163	26	.	.	PUNCT
ijassa-372	164	1	(	(	PUNCT
ijassa-372	164	2	a)traditional	a)traditional	ADJ
ijassa-372	164	3	bow	bow	NOUN
ijassa-372	164	4	(	(	PUNCT
ijassa-372	164	5	b)our	b)our	NUM
ijassa-372	164	6	method	method	NOUN
ijassa-372	164	7	fig.3	fig.3	PROPN
ijassa-372	164	8	retrieval	retrieval	NOUN
ijassa-372	164	9	example	example	NOUN
ijassa-372	164	10	for	for	ADP
ijassa-372	164	11	a	a	DET
ijassa-372	164	12	query	query	NOUN
ijassa-372	164	13	image	image	NOUN
ijassa-372	164	14	taken	take	VERB
ijassa-372	164	15	from	from	ADP
ijassa-372	164	16	the	the	DET
ijassa-372	164	17	“	"	PUNCT
ijassa-372	164	18	calsuburb	calsuburb	NOUN
ijassa-372	164	19	”	"	PUNCT
ijassa-372	164	20	category	category	NOUN
ijassa-372	164	21	.	.	PUNCT
ijassa-372	165	1	(	(	PUNCT
ijassa-372	165	2	a)traditional	a)traditional	ADJ
ijassa-372	165	3	bow	bow	NOUN
ijassa-372	165	4	(	(	PUNCT
ijassa-372	165	5	b)our	b)our	NUM
ijassa-372	165	6	method	method	NOUN
ijassa-372	165	7	fig.4	fig.4	PROPN
ijassa-372	165	8	retrieval	retrieval	NOUN
ijassa-372	165	9	example	example	NOUN
ijassa-372	165	10	for	for	ADP
ijassa-372	165	11	a	a	DET
ijassa-372	165	12	query	query	NOUN
ijassa-372	165	13	image	image	NOUN
ijassa-372	165	14	taken	take	VERB
ijassa-372	165	15	from	from	ADP
ijassa-372	165	16	the	the	DET
ijassa-372	165	17	“	"	PUNCT
ijassa-372	165	18	mittallbuilding	mittallbuilding	NOUN
ijassa-372	165	19	”	"	PUNCT
ijassa-372	165	20	category	category	NOUN
ijassa-372	165	21	.	.	PUNCT
ijassa-372	166	1	5	5	NUM
ijassa-372	166	2	conclusions	conclusion	NOUN
ijassa-372	166	3	in	in	ADP
ijassa-372	166	4	this	this	DET
ijassa-372	166	5	paper	paper	NOUN
ijassa-372	166	6	we	we	PRON
ijassa-372	166	7	present	present	VERB
ijassa-372	166	8	a	a	DET
ijassa-372	166	9	flexible	flexible	ADJ
ijassa-372	166	10	patch	patch	NOUN
ijassa-372	166	11	-	-	PUNCT
ijassa-372	166	12	based	base	VERB
ijassa-372	166	13	distance	distance	NOUN
ijassa-372	166	14	measure	measure	NOUN
ijassa-372	166	15	method	method	NOUN
ijassa-372	166	16	for	for	ADP
ijassa-372	166	17	images	image	NOUN
ijassa-372	166	18	,	,	PUNCT
ijassa-372	166	19	which	which	PRON
ijassa-372	166	20	utilizes	utilize	VERB
ijassa-372	166	21	the	the	DET
ijassa-372	166	22	abundant	abundant	ADJ
ijassa-372	166	23	local	local	ADJ
ijassa-372	166	24	feature	feature	NOUN
ijassa-372	166	25	of	of	ADP
ijassa-372	166	26	images	image	NOUN
ijassa-372	166	27	and	and	CCONJ
ijassa-372	166	28	the	the	DET
ijassa-372	166	29	spatial	spatial	ADJ
ijassa-372	166	30	information	information	NOUN
ijassa-372	166	31	of	of	ADP
ijassa-372	166	32	them	they	PRON
ijassa-372	166	33	.	.	PUNCT
ijassa-372	167	1	the	the	DET
ijassa-372	167	2	proposed	propose	VERB
ijassa-372	167	3	method	method	NOUN
ijassa-372	167	4	is	be	AUX
ijassa-372	167	5	tested	test	VERB
ijassa-372	167	6	in	in	ADP
ijassa-372	167	7	image	image	NOUN
ijassa-372	167	8	retrieval	retrieval	NOUN
ijassa-372	167	9	application	application	NOUN
ijassa-372	167	10	.	.	PUNCT
ijassa-372	168	1	experiments	experiment	NOUN
ijassa-372	168	2	show	show	VERB
ijassa-372	168	3	better	well	ADJ
ijassa-372	168	4	retrieval	retrieval	NOUN
ijassa-372	168	5	results	result	NOUN
ijassa-372	168	6	than	than	ADP
ijassa-372	168	7	the	the	DET
ijassa-372	168	8	traditional	traditional	ADJ
ijassa-372	168	9	bow	bow	NOUN
ijassa-372	168	10	.	.	PUNCT
ijassa-372	169	1	we	we	PRON
ijassa-372	169	2	use	use	VERB
ijassa-372	169	3	hard	hard	ADJ
ijassa-372	169	4	assignment	assignment	NOUN
ijassa-372	169	5	for	for	ADP
ijassa-372	169	6	testing	test	VERB
ijassa-372	169	7	our	our	PRON
ijassa-372	169	8	method	method	NOUN
ijassa-372	169	9	.	.	PUNCT
ijassa-372	170	1	in	in	ADP
ijassa-372	170	2	the	the	DET
ijassa-372	170	3	future	future	NOUN
ijassa-372	170	4	,	,	PUNCT
ijassa-372	170	5	we	we	PRON
ijassa-372	170	6	will	will	AUX
ijassa-372	170	7	test	test	VERB
ijassa-372	170	8	the	the	DET
ijassa-372	170	9	retrieval	retrieval	NOUN
ijassa-372	170	10	performance	performance	NOUN
ijassa-372	170	11	of	of	ADP
ijassa-372	170	12	integrating	integrate	VERB
ijassa-372	170	13	the	the	DET
ijassa-372	170	14	soft	soft	ADJ
ijassa-372	170	15	assignment	assignment	NOUN
ijassa-372	170	16	with	with	ADP
ijassa-372	170	17	our	our	PRON
ijassa-372	170	18	method	method	NOUN
ijassa-372	170	19	.	.	PUNCT
ijassa-372	171	1	in	in	ADP
ijassa-372	171	2	addition	addition	NOUN
ijassa-372	171	3	,	,	PUNCT
ijassa-372	171	4	we	we	PRON
ijassa-372	171	5	only	only	ADV
ijassa-372	171	6	utilize	utilize	VERB
ijassa-372	171	7	the	the	DET
ijassa-372	171	8	absolute	absolute	ADJ
ijassa-372	171	9	position	position	NOUN
ijassa-372	171	10	of	of	ADP
ijassa-372	171	11	visual	visual	ADJ
ijassa-372	171	12	words	word	NOUN
ijassa-372	171	13	in	in	ADP
ijassa-372	171	14	our	our	PRON
ijassa-372	171	15	method	method	NOUN
ijassa-372	171	16	.	.	PUNCT
ijassa-372	172	1	the	the	DET
ijassa-372	172	2	relative	relative	ADJ
ijassa-372	172	3	relationship	relationship	NOUN
ijassa-372	172	4	of	of	ADP
ijassa-372	172	5	visual	visual	ADJ
ijassa-372	172	6	words	word	NOUN
ijassa-372	172	7	is	be	AUX
ijassa-372	172	8	more	more	ADV
ijassa-372	172	9	robust	robust	ADJ
ijassa-372	172	10	to	to	ADP
ijassa-372	172	11	image	image	NOUN
ijassa-372	172	12	transformation	transformation	NOUN
ijassa-372	172	13	,	,	PUNCT
ijassa-372	172	14	which	which	PRON
ijassa-372	172	15	will	will	AUX
ijassa-372	172	16	advances	advance	VERB
ijassa-372	172	17	in	in	ADP
ijassa-372	172	18	systems	system	NOUN
ijassa-372	172	19	science	science	NOUN
ijassa-372	172	20	and	and	CCONJ
ijassa-372	172	21	application	application	NOUN
ijassa-372	172	22	(	(	PUNCT
ijassa-372	172	23	2015	2015	NUM
ijassa-372	172	24	)	)	PUNCT
ijassa-372	172	25	vol.15	vol.15	NOUN
ijassa-372	172	26	no.3	no.3	VERB
ijassa-372	172	27	251	251	NUM
ijassa-372	172	28	be	be	AUX
ijassa-372	172	29	our	our	PRON
ijassa-372	172	30	next	next	ADJ
ijassa-372	172	31	research	research	NOUN
ijassa-372	172	32	work	work	NOUN
ijassa-372	172	33	.	.	PUNCT
ijassa-372	173	1	(	(	PUNCT
ijassa-372	173	2	a)traditional	a)traditional	ADJ
ijassa-372	173	3	bow	bow	NOUN
ijassa-372	173	4	(	(	PUNCT
ijassa-372	173	5	b)our	b)our	NUM
ijassa-372	173	6	method	method	NOUN
ijassa-372	173	7	fig.5	fig.5	VERB
ijassa-372	173	8	retrieval	retrieval	NOUN
ijassa-372	173	9	example	example	NOUN
ijassa-372	173	10	for	for	ADP
ijassa-372	173	11	a	a	DET
ijassa-372	173	12	query	query	NOUN
ijassa-372	173	13	image	image	NOUN
ijassa-372	173	14	taken	take	VERB
ijassa-372	173	15	from	from	ADP
ijassa-372	173	16	the	the	DET
ijassa-372	173	17	“	"	PUNCT
ijassa-372	173	18	mitmountain	mitmountain	NOUN
ijassa-372	173	19	”	"	PUNCT
ijassa-372	173	20	category	category	NOUN
ijassa-372	173	21	.	.	PUNCT
ijassa-372	174	1	acknowledgments	acknowledgment	NOUN
ijassa-372	174	2	this	this	DET
ijassa-372	174	3	work	work	NOUN
ijassa-372	174	4	was	be	AUX
ijassa-372	174	5	supported	support	VERB
ijassa-372	174	6	by	by	ADP
ijassa-372	174	7	hubei	hubei	PROPN
ijassa-372	174	8	province	province	PROPN
ijassa-372	174	9	key	key	PROPN
ijassa-372	174	10	laboratory	laboratory	NOUN
ijassa-372	174	11	of	of	ADP
ijassa-372	174	12	systems	system	NOUN
ijassa-372	174	13	science	science	NOUN
ijassa-372	174	14	in	in	ADP
ijassa-372	174	15	metallurgical	metallurgical	ADJ
ijassa-372	174	16	process	process	NOUN
ijassa-372	174	17	(	(	PUNCT
ijassa-372	174	18	wuhan	wuhan	PROPN
ijassa-372	174	19	university	university	PROPN
ijassa-372	174	20	of	of	ADP
ijassa-372	174	21	science	science	NOUN
ijassa-372	174	22	and	and	CCONJ
ijassa-372	174	23	technology):y201406	technology):y201406	NUM
ijassa-372	174	24	,	,	PUNCT
ijassa-372	174	25	and	and	CCONJ
ijassa-372	174	26	the	the	DET
ijassa-372	174	27	natural	natural	ADJ
ijassa-372	174	28	science	science	NOUN
ijassa-372	174	29	foundation	foundation	PROPN
ijassa-372	174	30	of	of	ADP
ijassa-372	174	31	hubei	hubei	PROPN
ijassa-372	174	32	province	province	PROPN
ijassa-372	174	33	:	:	PUNCT
ijassa-372	175	1	22013cfa131	22013cfa131	X
ijassa-372	175	2	.	.	PUNCT
ijassa-372	175	3	references	reference	NOUN
ijassa-372	175	4	[	[	X
ijassa-372	175	5	1	1	X
ijassa-372	175	6	]	]	PUNCT
ijassa-372	175	7	j.	j.	PROPN
ijassa-372	175	8	philbin	philbin	PROPN
ijassa-372	175	9	,	,	PUNCT
ijassa-372	175	10	o.	o.	NOUN
ijassa-372	175	11	chum	chum	NOUN
ijassa-372	175	12	,	,	PUNCT
ijassa-372	175	13	m.	m.	NOUN
ijassa-372	175	14	isard	isard	NOUN
ijassa-372	175	15	,	,	PUNCT
ijassa-372	175	16	et	et	PROPN
ijassa-372	175	17	al	al	PROPN
ijassa-372	175	18	.	.	PUNCT
ijassa-372	175	19	(	(	PUNCT
ijassa-372	175	20	2007	2007	NUM
ijassa-372	175	21	)	)	PUNCT
ijassa-372	175	22	,	,	PUNCT
ijassa-372	175	23	“	"	PUNCT
ijassa-372	175	24	object	object	VERB
ijassa-372	175	25	retrieval	retrieval	NOUN
ijassa-372	175	26	with	with	ADP
ijassa-372	175	27	large	large	ADJ
ijassa-372	175	28	vocabularies	vocabulary	NOUN
ijassa-372	175	29	and	and	CCONJ
ijassa-372	175	30	fast	fast	ADJ
ijassa-372	175	31	spatial	spatial	ADJ
ijassa-372	175	32	matching	matching	NOUN
ijassa-372	175	33	”	"	PUNCT
ijassa-372	175	34	,	,	PUNCT
ijassa-372	175	35	ieee	ieee	NOUN
ijassa-372	175	36	conference	conference	NOUN
ijassa-372	175	37	on	on	ADP
ijassa-372	175	38	computer	computer	NOUN
ijassa-372	175	39	vision	vision	NOUN
ijassa-372	175	40	and	and	CCONJ
ijassa-372	175	41	pattern	pattern	NOUN
ijassa-372	175	42	recognition	recognition	NOUN
ijassa-372	175	43	.	.	PUNCT
ijassa-372	176	1	[	[	X
ijassa-372	176	2	2	2	X
ijassa-372	176	3	]	]	PUNCT
ijassa-372	176	4	w.	w.	PROPN
ijassa-372	176	5	zhou	zhou	PROPN
ijassa-372	176	6	,	,	PUNCT
ijassa-372	176	7	y.	y.	PROPN
ijassa-372	176	8	lu	lu	PROPN
ijassa-372	176	9	,	,	PUNCT
ijassa-372	176	10	h.	h.	PROPN
ijassa-372	176	11	li	li	PROPN
ijassa-372	176	12	,	,	PUNCT
ijassa-372	176	13	et	et	NOUN
ijassa-372	176	14	al.(2010	al.(2010	NOUN
ijassa-372	176	15	)	)	PUNCT
ijassa-372	176	16	,	,	PUNCT
ijassa-372	176	17	“	"	PUNCT
ijassa-372	176	18	spatial	spatial	ADJ
ijassa-372	176	19	coding	coding	NOUN
ijassa-372	176	20	for	for	ADP
ijassa-372	176	21	large	large	ADJ
ijassa-372	176	22	scale	scale	NOUN
ijassa-372	176	23	partialduplicate	partialduplicate	NOUN
ijassa-372	176	24	web	web	NOUN
ijassa-372	176	25	image	image	NOUN
ijassa-372	176	26	search	search	NOUN
ijassa-372	176	27	”	"	PUNCT
ijassa-372	176	28	,	,	PUNCT
ijassa-372	176	29	18th	18th	ADJ
ijassa-372	176	30	acm	acm	PROPN
ijassa-372	176	31	international	international	ADJ
ijassa-372	176	32	conference	conference	NOUN
ijassa-372	176	33	on	on	ADP
ijassa-372	176	34	multimedia	multimedia	NOUN
ijassa-372	176	35	,	,	PUNCT
ijassa-372	176	36	pp.511	pp.511	NOUN
ijassa-372	176	37	-	-	PUNCT
ijassa-372	176	38	520	520	NUM
ijassa-372	176	39	.	.	PUNCT
ijassa-372	177	1	[	[	X
ijassa-372	177	2	3	3	X
ijassa-372	177	3	]	]	X
ijassa-372	177	4	h.	h.	PROPN
ijassa-372	177	5	jégou	jégou	PROPN
ijassa-372	177	6	,	,	PUNCT
ijassa-372	177	7	m.	m.	NOUN
ijassa-372	177	8	douze	douze	PROPN
ijassa-372	177	9	and	and	CCONJ
ijassa-372	177	10	c.	c.	PROPN
ijassa-372	177	11	schmid.(2010	schmid.(2010	PROPN
ijassa-372	177	12	)	)	PUNCT
ijassa-372	177	13	,	,	PUNCT
ijassa-372	177	14	“	"	PUNCT
ijassa-372	177	15	improving	improve	VERB
ijassa-372	177	16	bag	bag	NOUN
ijassa-372	177	17	-	-	PUNCT
ijassa-372	177	18	of	of	ADP
ijassa-372	177	19	-	-	PUNCT
ijassa-372	177	20	features	feature	NOUN
ijassa-372	177	21	for	for	ADP
ijassa-372	177	22	large	large	ADJ
ijassa-372	177	23	scale	scale	NOUN
ijassa-372	177	24	image	image	NOUN
ijassa-372	177	25	search	search	NOUN
ijassa-372	177	26	”	"	PUNCT
ijassa-372	177	27	international	international	ADJ
ijassa-372	177	28	journal	journal	NOUN
ijassa-372	177	29	of	of	ADP
ijassa-372	177	30	computer	computer	NOUN
ijassa-372	177	31	vision	vision	NOUN
ijassa-372	177	32	,	,	PUNCT
ijassa-372	177	33	vol.87,pp.316	vol.87,pp.316	PROPN
ijassa-372	177	34	-	-	NOUN
ijassa-372	177	35	336	336	NUM
ijassa-372	177	36	.	.	PUNCT
ijassa-372	178	1	[	[	X
ijassa-372	178	2	4	4	X
ijassa-372	178	3	]	]	PUNCT
ijassa-372	178	4	s.	s.	PROPN
ijassa-372	178	5	lazebnik	lazebnik	PROPN
ijassa-372	178	6	,	,	PUNCT
ijassa-372	178	7	c.	c.	PROPN
ijassa-372	178	8	schmid	schmid	PROPN
ijassa-372	178	9	and	and	CCONJ
ijassa-372	178	10	j.	j.	PROPN
ijassa-372	178	11	ponce	ponce	PROPN
ijassa-372	178	12	.	.	PUNCT
ijassa-372	179	1	(	(	PUNCT
ijassa-372	179	2	2006	2006	NUM
ijassa-372	179	3	)	)	PUNCT
ijassa-372	179	4	,	,	PUNCT
ijassa-372	179	5	“	"	PUNCT
ijassa-372	179	6	beyond	beyond	ADP
ijassa-372	179	7	bags	bag	NOUN
ijassa-372	179	8	of	of	ADP
ijassa-372	179	9	features	feature	NOUN
ijassa-372	179	10	:	:	PUNCT
ijassa-372	179	11	spatial	spatial	ADJ
ijassa-372	179	12	pyramid	pyramid	NOUN
ijassa-372	179	13	matching	match	VERB
ijassa-372	179	14	for	for	ADP
ijassa-372	179	15	recognizing	recognize	VERB
ijassa-372	179	16	natural	natural	ADJ
ijassa-372	179	17	scene	scene	NOUN
ijassa-372	179	18	categories	category	NOUN
ijassa-372	179	19	”	"	PUNCT
ijassa-372	179	20	,	,	PUNCT
ijassa-372	179	21	ieee	ieee	NOUN
ijassa-372	179	22	conference	conference	NOUN
ijassa-372	179	23	on	on	ADP
ijassa-372	179	24	computer	computer	NOUN
ijassa-372	179	25	vision	vision	NOUN
ijassa-372	179	26	and	and	CCONJ
ijassa-372	179	27	pattern	pattern	NOUN
ijassa-372	179	28	recognition	recognition	NOUN
ijassa-372	179	29	,	,	PUNCT
ijassa-372	179	30	pp.2169	pp.2169	PROPN
ijassa-372	179	31	-	-	PUNCT
ijassa-372	179	32	2178	2178	NUM
ijassa-372	179	33	.	.	PUNCT
ijassa-372	180	1	[	[	X
ijassa-372	180	2	5	5	X
ijassa-372	180	3	]	]	PUNCT
ijassa-372	180	4	j.	j.	PROPN
ijassa-372	180	5	yang	yang	PROPN
ijassa-372	180	6	,	,	PUNCT
ijassa-372	180	7	k.	k.	PROPN
ijassa-372	180	8	yu	yu	PROPN
ijassa-372	180	9	,	,	PUNCT
ijassa-372	180	10	y.	y.	PROPN
ijassa-372	180	11	gong	gong	PROPN
ijassa-372	180	12	et	et	PROPN
ijassa-372	180	13	al	al	PROPN
ijassa-372	180	14	.	.	PROPN
ijassa-372	180	15	(	(	PUNCT
ijassa-372	180	16	2009	2009	NUM
ijassa-372	180	17	)	)	PUNCT
ijassa-372	180	18	,	,	PUNCT
ijassa-372	180	19	“	"	PUNCT
ijassa-372	180	20	linear	linear	ADJ
ijassa-372	180	21	spatial	spatial	ADJ
ijassa-372	180	22	pyramid	pyramid	NOUN
ijassa-372	180	23	matching	match	VERB
ijassa-372	180	24	using	use	VERB
ijassa-372	180	25	sparse	sparse	ADJ
ijassa-372	180	26	coding	coding	NOUN
ijassa-372	180	27	for	for	ADP
ijassa-372	180	28	image	image	NOUN
ijassa-372	180	29	classification	classification	NOUN
ijassa-372	180	30	”	"	PUNCT
ijassa-372	180	31	,	,	PUNCT
ijassa-372	180	32	ieee	ieee	NOUN
ijassa-372	180	33	conference	conference	NOUN
ijassa-372	180	34	on	on	ADP
ijassa-372	180	35	computer	computer	NOUN
ijassa-372	180	36	vision	vision	NOUN
ijassa-372	180	37	and	and	CCONJ
ijassa-372	180	38	pattern	pattern	NOUN
ijassa-372	180	39	recognition	recognition	NOUN
ijassa-372	180	40	,	,	PUNCT
ijassa-372	180	41	pp.1794	pp.1794	PROPN
ijassa-372	180	42	-	-	PUNCT
ijassa-372	180	43	1801	1801	NUM
ijassa-372	180	44	.	.	PUNCT
ijassa-372	181	1	252	252	NUM
ijassa-372	181	2	yan	yan	PROPN
ijassa-372	181	3	yu	yu	PROPN
ijassa-372	181	4	:	:	PUNCT
ijassa-372	181	5	image	image	NOUN
ijassa-372	181	6	retrieval	retrieval	NOUN
ijassa-372	181	7	based	base	VERB
ijassa-372	181	8	on	on	ADP
ijassa-372	181	9	spatial	spatial	ADJ
ijassa-372	181	10	organization	organization	NOUN
ijassa-372	181	11	of	of	ADP
ijassa-372	181	12	patches	patch	NOUN
ijassa-372	181	13	[	[	X
ijassa-372	181	14	6	6	NUM
ijassa-372	181	15	]	]	PUNCT
ijassa-372	181	16	c.	c.	PROPN
ijassa-372	181	17	c.	c.	PROPN
ijassa-372	181	18	yan	yan	PROPN
ijassa-372	181	19	,	,	PUNCT
ijassa-372	181	20	l.	l.	PROPN
ijassa-372	181	21	li	li	PROPN
ijassa-372	181	22	,	,	PUNCT
ijassa-372	181	23	z.	z.	PROPN
ijassa-372	181	24	wang	wang	PROPN
ijassa-372	181	25	,	,	PUNCT
ijassa-372	181	26	et	et	PROPN
ijassa-372	181	27	al	al	PROPN
ijassa-372	181	28	.	.	PUNCT
ijassa-372	181	29	(	(	PUNCT
ijassa-372	181	30	2014	2014	NUM
ijassa-372	181	31	)	)	PUNCT
ijassa-372	181	32	,	,	PUNCT
ijassa-372	181	33	“	"	PUNCT
ijassa-372	181	34	fusing	fuse	VERB
ijassa-372	181	35	multi	multi	ADJ
ijassa-372	181	36	-	-	ADJ
ijassa-372	181	37	cues	cues	ADJ
ijassa-372	181	38	description	description	NOUN
ijassa-372	181	39	for	for	ADP
ijassa-372	181	40	partial	partial	ADJ
ijassa-372	181	41	-	-	PUNCT
ijassa-372	181	42	duplicate	duplicate	ADJ
ijassa-372	181	43	image	image	NOUN
ijassa-372	181	44	retrieval	retrieval	NOUN
ijassa-372	181	45	”	"	PUNCT
ijassa-372	181	46	journal	journal	NOUN
ijassa-372	181	47	of	of	ADP
ijassa-372	181	48	visual	visual	ADJ
ijassa-372	181	49	communication	communication	NOUN
ijassa-372	181	50	and	and	CCONJ
ijassa-372	181	51	image	image	NOUN
ijassa-372	181	52	representation	representation	NOUN
ijassa-372	181	53	,	,	PUNCT
ijassa-372	181	54	vol.25	vol.25	ADJ
ijassa-372	181	55	,	,	PUNCT
ijassa-372	181	56	no.6	no.6	PROPN
ijassa-372	181	57	,	,	PUNCT
ijassa-372	181	58	pp	pp	ADJ
ijassa-372	181	59	.	.	PUNCT
ijassa-372	181	60	1726	1726	NUM
ijassa-372	181	61	-	-	SYM
ijassa-372	181	62	1731	1731	NUM
ijassa-372	181	63	.	.	PUNCT
ijassa-372	182	1	[	[	X
ijassa-372	182	2	7	7	X
ijassa-372	182	3	]	]	X
ijassa-372	182	4	t.	t.	PROPN
ijassa-372	182	5	hurtut	hurtut	PROPN
ijassa-372	182	6	,	,	PUNCT
ijassa-372	182	7	y.	y.	NOUN
ijassa-372	182	8	gousseau	gousseau	PROPN
ijassa-372	182	9	and	and	CCONJ
ijassa-372	182	10	f.	f.	PROPN
ijassa-372	182	11	schmitt	schmitt	PROPN
ijassa-372	182	12	.	.	PUNCT
ijassa-372	183	1	(	(	PUNCT
ijassa-372	183	2	2008	2008	NUM
ijassa-372	183	3	)	)	PUNCT
ijassa-372	183	4	,	,	PUNCT
ijassa-372	183	5	“	"	PUNCT
ijassa-372	183	6	adaptive	adaptive	ADJ
ijassa-372	183	7	image	image	NOUN
ijassa-372	183	8	retrieval	retrieval	NOUN
ijassa-372	183	9	based	base	VERB
ijassa-372	183	10	on	on	ADP
ijassa-372	183	11	the	the	DET
ijassa-372	183	12	spatial	spatial	ADJ
ijassa-372	183	13	organization	organization	NOUN
ijassa-372	183	14	of	of	ADP
ijassa-372	183	15	colors	color	NOUN
ijassa-372	183	16	”	"	PUNCT
ijassa-372	183	17	,	,	PUNCT
ijassa-372	183	18	computer	computer	NOUN
ijassa-372	183	19	vision	vision	NOUN
ijassa-372	183	20	and	and	CCONJ
ijassa-372	183	21	image	image	NOUN
ijassa-372	183	22	understanding	understanding	NOUN
ijassa-372	183	23	,	,	PUNCT
ijassa-372	183	24	vol.112	vol.112	NOUN
ijassa-372	183	25	,	,	PUNCT
ijassa-372	183	26	pp.101	pp.101	NOUN
ijassa-372	183	27	-	-	PUNCT
ijassa-372	183	28	113	113	NUM
ijassa-372	183	29	.	.	PUNCT
ijassa-372	184	1	[	[	X
ijassa-372	184	2	8	8	X
ijassa-372	184	3	]	]	PUNCT
ijassa-372	184	4	j.	j.	PROPN
ijassa-372	184	5	huang	huang	PROPN
ijassa-372	184	6	,	,	PUNCT
ijassa-372	184	7	s.	s.	PROPN
ijassa-372	184	8	r.	r.	PROPN
ijassa-372	184	9	kumar	kumar	PROPN
ijassa-372	184	10	,	,	PUNCT
ijassa-372	184	11	m.	m.	PROPN
ijassa-372	184	12	mitra	mitra	PROPN
ijassa-372	184	13	,	,	PUNCT
ijassa-372	184	14	et	et	PROPN
ijassa-372	184	15	al	al	PROPN
ijassa-372	184	16	.	.	PROPN
ijassa-372	184	17	(	(	PUNCT
ijassa-372	184	18	1997	1997	NUM
ijassa-372	184	19	)	)	PUNCT
ijassa-372	184	20	,	,	PUNCT
ijassa-372	184	21	“	"	PUNCT
ijassa-372	184	22	image	image	NOUN
ijassa-372	184	23	indexing	indexing	NOUN
ijassa-372	184	24	using	use	VERB
ijassa-372	184	25	color	color	NOUN
ijassa-372	184	26	correlograms	correlogram	NOUN
ijassa-372	184	27	”	"	PUNCT
ijassa-372	184	28	,	,	PUNCT
ijassa-372	184	29	ieee	ieee	NOUN
ijassa-372	184	30	conference	conference	NOUN
ijassa-372	184	31	on	on	ADP
ijassa-372	184	32	computer	computer	NOUN
ijassa-372	184	33	vision	vision	NOUN
ijassa-372	184	34	and	and	CCONJ
ijassa-372	184	35	pattern	pattern	NOUN
ijassa-372	184	36	recognition	recognition	NOUN
ijassa-372	184	37	,	,	PUNCT
ijassa-372	184	38	pp.762	pp.762	PROPN
ijassa-372	184	39	-	-	PUNCT
ijassa-372	184	40	768	768	NUM
ijassa-372	184	41	.	.	PUNCT
ijassa-372	185	1	[	[	X
ijassa-372	185	2	9	9	NUM
ijassa-372	185	3	]	]	PUNCT
ijassa-372	185	4	h.	h.	PROPN
ijassa-372	185	5	e.	e.	PROPN
ijassa-372	185	6	j.	j.	PROPN
ijassa-372	185	7	egou	egou	PROPN
ijassa-372	185	8	,	,	PUNCT
ijassa-372	185	9	m.	m.	NOUN
ijassa-372	185	10	douze	douze	PROPN
ijassa-372	185	11	and	and	CCONJ
ijassa-372	185	12	c.	c.	PROPN
ijassa-372	185	13	schmid	schmid	PROPN
ijassa-372	185	14	.	.	PUNCT
ijassa-372	186	1	(	(	PUNCT
ijassa-372	186	2	2008	2008	NUM
ijassa-372	186	3	)	)	PUNCT
ijassa-372	186	4	,	,	PUNCT
ijassa-372	186	5	“	"	PUNCT
ijassa-372	186	6	hamming	ham	VERB
ijassa-372	186	7	embedding	embed	VERB
ijassa-372	186	8	and	and	CCONJ
ijassa-372	186	9	weak	weak	ADJ
ijassa-372	186	10	geometric	geometric	ADJ
ijassa-372	186	11	consistency	consistency	NOUN
ijassa-372	186	12	for	for	ADP
ijassa-372	186	13	large	large	ADJ
ijassa-372	186	14	scale	scale	NOUN
ijassa-372	186	15	image	image	NOUN
ijassa-372	186	16	search	search	NOUN
ijassa-372	186	17	”	"	PUNCT
ijassa-372	186	18	,	,	PUNCT
ijassa-372	186	19	10th	10th	ADJ
ijassa-372	186	20	european	european	ADJ
ijassa-372	186	21	conference	conference	NOUN
ijassa-372	186	22	on	on	ADP
ijassa-372	186	23	computer	computer	NOUN
ijassa-372	186	24	vision	vision	NOUN
ijassa-372	186	25	,	,	PUNCT
ijassa-372	186	26	pp.304	pp.304	PROPN
ijassa-372	186	27	-	-	PUNCT
ijassa-372	186	28	317	317	NUM
ijassa-372	186	29	.	.	PUNCT
ijassa-372	187	1	[	[	X
ijassa-372	187	2	10	10	NUM
ijassa-372	187	3	]	]	X
ijassa-372	187	4	o.	o.	NOUN
ijassa-372	187	5	chum	chum	NOUN
ijassa-372	187	6	,	,	PUNCT
ijassa-372	187	7	m.	m.	NOUN
ijassa-372	187	8	perdoch	perdoch	PROPN
ijassa-372	187	9	and	and	CCONJ
ijassa-372	187	10	j.	j.	PROPN
ijassa-372	187	11	matas	matas	PROPN
ijassa-372	187	12	.	.	PUNCT
ijassa-372	187	13	(	(	PUNCT
ijassa-372	187	14	2009	2009	NUM
ijassa-372	187	15	)	)	PUNCT
ijassa-372	187	16	,	,	PUNCT
ijassa-372	187	17	“	"	PUNCT
ijassa-372	187	18	geometric	geometric	ADJ
ijassa-372	187	19	min	min	NOUN
ijassa-372	187	20	-	-	ADJ
ijassa-372	187	21	hashing	hashing	NOUN
ijassa-372	187	22	:	:	PUNCT
ijassa-372	187	23	finding	find	VERB
ijassa-372	187	24	a	a	DET
ijassa-372	187	25	(	(	PUNCT
ijassa-372	187	26	thick	thick	ADJ
ijassa-372	187	27	)	)	PUNCT
ijassa-372	187	28	needle	needle	NOUN
ijassa-372	187	29	in	in	ADP
ijassa-372	187	30	a	a	DET
ijassa-372	187	31	haystack	haystack	NOUN
ijassa-372	187	32	”	"	PUNCT
ijassa-372	187	33	,	,	PUNCT
ijassa-372	187	34	ieee	ieee	NOUN
ijassa-372	187	35	conference	conference	NOUN
ijassa-372	187	36	on	on	ADP
ijassa-372	187	37	computer	computer	NOUN
ijassa-372	187	38	vision	vision	NOUN
ijassa-372	187	39	and	and	CCONJ
ijassa-372	187	40	pattern	pattern	NOUN
ijassa-372	187	41	recognition	recognition	NOUN
ijassa-372	187	42	,	,	PUNCT
ijassa-372	187	43	pp.17	pp.17	PROPN
ijassa-372	187	44	-	-	SYM
ijassa-372	187	45	24	24	NUM
ijassa-372	187	46	.	.	PUNCT
ijassa-372	188	1	[	[	X
ijassa-372	188	2	11	11	NUM
ijassa-372	188	3	]	]	PUNCT
ijassa-372	188	4	j.	j.	PROPN
ijassa-372	188	5	feng	feng	PROPN
ijassa-372	188	6	,	,	PUNCT
ijassa-372	188	7	b.	b.	PROPN
ijassa-372	188	8	ni	ni	PROPN
ijassa-372	188	9	,	,	PUNCT
ijassa-372	188	10	q.	q.	PROPN
ijassa-372	188	11	tian	tian	PROPN
ijassa-372	188	12	,	,	PUNCT
ijassa-372	188	13	et	et	PROPN
ijassa-372	188	14	al	al	PROPN
ijassa-372	188	15	.	.	PUNCT
ijassa-372	188	16	(	(	PUNCT
ijassa-372	188	17	2011	2011	NUM
ijassa-372	188	18	)	)	PUNCT
ijassa-372	188	19	,	,	PUNCT
ijassa-372	188	20	“	"	PUNCT
ijassa-372	188	21	geometric	geometric	ADJ
ijassa-372	188	22	lp	lp	ADJ
ijassa-372	188	23	-	-	PUNCT
ijassa-372	188	24	norm	norm	NOUN
ijassa-372	188	25	feature	feature	NOUN
ijassa-372	188	26	pooling	pool	VERB
ijassa-372	188	27	for	for	ADP
ijassa-372	188	28	image	image	NOUN
ijassa-372	188	29	classification”,ieee	classification”,ieee	NOUN
ijassa-372	188	30	conference	conference	NOUN
ijassa-372	188	31	on	on	ADP
ijassa-372	188	32	computer	computer	NOUN
ijassa-372	188	33	vision	vision	NOUN
ijassa-372	188	34	and	and	CCONJ
ijassa-372	188	35	pattern	pattern	NOUN
ijassa-372	188	36	recognition	recognition	NOUN
ijassa-372	188	37	,	,	PUNCT
ijassa-372	188	38	pp.2609	pp.2609	PROPN
ijassa-372	188	39	-	-	PUNCT
ijassa-372	188	40	2704	2704	NUM
ijassa-372	188	41	.	.	PUNCT
ijassa-372	189	1	[	[	X
ijassa-372	189	2	12	12	NUM
ijassa-372	189	3	]	]	X
ijassa-372	189	4	n.	n.	NOUN
ijassa-372	189	5	v.	v.	ADP
ijassa-372	189	6	hoàng	hoàng	PROPN
ijassa-372	189	7	,	,	PUNCT
ijassa-372	189	8	v.	v.	ADP
ijassa-372	189	9	gouet	gouet	NOUN
ijassa-372	189	10	-	-	PUNCT
ijassa-372	189	11	brunet	brunet	NOUN
ijassa-372	189	12	,	,	PUNCT
ijassa-372	189	13	m.	m.	NOUN
ijassa-372	189	14	rukoz	rukoz	NOUN
ijassa-372	189	15	,	,	PUNCT
ijassa-372	189	16	et	et	PROPN
ijassa-372	189	17	al	al	PROPN
ijassa-372	189	18	.	.	PUNCT
ijassa-372	190	1	(	(	PUNCT
ijassa-372	190	2	2010	2010	NUM
ijassa-372	190	3	)	)	PUNCT
ijassa-372	190	4	,	,	PUNCT
ijassa-372	190	5	“	"	PUNCT
ijassa-372	190	6	embedding	embed	VERB
ijassa-372	190	7	spatial	spatial	ADJ
ijassa-372	190	8	information	information	NOUN
ijassa-372	190	9	into	into	ADP
ijassa-372	190	10	image	image	NOUN
ijassa-372	190	11	content	content	NOUN
ijassa-372	190	12	description	description	NOUN
ijassa-372	190	13	for	for	ADP
ijassa-372	190	14	scene	scene	NOUN
ijassa-372	190	15	retrieval	retrieval	NOUN
ijassa-372	190	16	”	"	PUNCT
ijassa-372	190	17	,	,	PUNCT
ijassa-372	190	18	pattern	pattern	NOUN
ijassa-372	190	19	recognition	recognition	NOUN
ijassa-372	190	20	,	,	PUNCT
ijassa-372	190	21	vol.43	vol.43	NOUN
ijassa-372	190	22	,	,	PUNCT
ijassa-372	190	23	pp.3013	pp.3013	PROPN
ijassa-372	190	24	-	-	PUNCT
ijassa-372	190	25	3024	3024	NUM
ijassa-372	190	26	.	.	PUNCT
ijassa-372	191	1	[	[	X
ijassa-372	191	2	13	13	NUM
ijassa-372	191	3	]	]	X
ijassa-372	191	4	s.	s.	PROPN
ijassa-372	191	5	karaman	karaman	PROPN
ijassa-372	191	6	,	,	PUNCT
ijassa-372	191	7	j.	j.	PROPN
ijassa-372	191	8	benois	benois	PROPN
ijassa-372	191	9	-	-	PUNCT
ijassa-372	191	10	pineau	pineau	PROPN
ijassa-372	191	11	,	,	PUNCT
ijassa-372	191	12	r.	r.	PROPN
ijassa-372	191	13	megret	megret	PROPN
ijassa-372	191	14	,	,	PUNCT
ijassa-372	191	15	et	et	PROPN
ijassa-372	191	16	al	al	PROPN
ijassa-372	191	17	.	.	PUNCT
ijassa-372	191	18	(	(	PUNCT
ijassa-372	191	19	2012	2012	NUM
ijassa-372	191	20	)	)	PUNCT
ijassa-372	191	21	,	,	PUNCT
ijassa-372	191	22	“	"	PUNCT
ijassa-372	191	23	multi	multi	ADJ
ijassa-372	191	24	-	-	ADJ
ijassa-372	191	25	layer	layer	ADJ
ijassa-372	191	26	local	local	ADJ
ijassa-372	191	27	graph	graph	NOUN
ijassa-372	191	28	words	word	NOUN
ijassa-372	191	29	for	for	ADP
ijassa-372	191	30	object	object	NOUN
ijassa-372	191	31	recognition	recognition	NOUN
ijassa-372	191	32	”	"	PUNCT
ijassa-372	191	33	,	,	PUNCT
ijassa-372	191	34	18th	18th	ADJ
ijassa-372	191	35	international	international	ADJ
ijassa-372	191	36	conference	conference	NOUN
ijassa-372	191	37	on	on	ADP
ijassa-372	191	38	multimedia	multimedia	NOUN
ijassa-372	191	39	modeling	modeling	NOUN
ijassa-372	191	40	,	,	PUNCT
ijassa-372	191	41	pp.29	pp.29	NOUN
ijassa-372	191	42	-	-	PUNCT
ijassa-372	191	43	39	39	NUM
ijassa-372	191	44	.	.	PUNCT
ijassa-372	192	1	[	[	X
ijassa-372	192	2	14	14	NUM
ijassa-372	192	3	]	]	X
ijassa-372	192	4	d.	d.	PROPN
ijassa-372	192	5	g.	g.	PROPN
ijassa-372	192	6	lowe	lowe	PROPN
ijassa-372	192	7	.	.	PUNCT
ijassa-372	193	1	(	(	PUNCT
ijassa-372	193	2	1999	1999	NUM
ijassa-372	193	3	)	)	PUNCT
ijassa-372	194	1	,	,	PUNCT
ijassa-372	194	2	“	"	PUNCT
ijassa-372	194	3	object	object	VERB
ijassa-372	194	4	recognition	recognition	NOUN
ijassa-372	194	5	from	from	ADP
ijassa-372	194	6	local	local	ADJ
ijassa-372	194	7	scale	scale	NOUN
ijassa-372	194	8	-	-	PUNCT
ijassa-372	194	9	invariant	invariant	ADJ
ijassa-372	194	10	features	feature	NOUN
ijassa-372	194	11	”	"	PUNCT
ijassa-372	194	12	,	,	PUNCT
ijassa-372	194	13	7th	7th	ADJ
ijassa-372	194	14	ieee	ieee	PROPN
ijassa-372	194	15	international	international	ADJ
ijassa-372	194	16	conference	conference	NOUN
ijassa-372	194	17	on	on	ADP
ijassa-372	194	18	computer	computer	NOUN
ijassa-372	194	19	vision	vision	NOUN
ijassa-372	194	20	,	,	PUNCT
ijassa-372	194	21	pp.1150	pp.1150	PROPN
ijassa-372	194	22	-	-	PUNCT
ijassa-372	194	23	1157	1157	NUM
ijassa-372	194	24	.	.	PUNCT
ijassa-372	195	1	[	[	X
ijassa-372	195	2	15	15	NUM
ijassa-372	195	3	]	]	X
ijassa-372	195	4	k.	k.	NOUN
ijassa-372	195	5	mikolajczyk	mikolajczyk	PROPN
ijassa-372	195	6	and	and	CCONJ
ijassa-372	195	7	c.	c.	PROPN
ijassa-372	195	8	schmid	schmid	PROPN
ijassa-372	195	9	.	.	PUNCT
ijassa-372	196	1	(	(	PUNCT
ijassa-372	196	2	2005	2005	NUM
ijassa-372	196	3	)	)	PUNCT
ijassa-372	196	4	,	,	PUNCT
ijassa-372	196	5	“	"	PUNCT
ijassa-372	196	6	a	a	DET
ijassa-372	196	7	performance	performance	NOUN
ijassa-372	196	8	evaluation	evaluation	NOUN
ijassa-372	196	9	of	of	ADP
ijassa-372	196	10	local	local	ADJ
ijassa-372	196	11	descriptors	descriptor	NOUN
ijassa-372	196	12	”	"	PUNCT
ijassa-372	196	13	,	,	PUNCT
ijassa-372	196	14	ieee	ieee	NOUN
ijassa-372	196	15	transactions	transaction	NOUN
ijassa-372	196	16	on	on	ADP
ijassa-372	196	17	pattern	pattern	NOUN
ijassa-372	196	18	analysis	analysis	NOUN
ijassa-372	196	19	and	and	CCONJ
ijassa-372	196	20	machine	machine	NOUN
ijassa-372	196	21	intelligence	intelligence	NOUN
ijassa-372	196	22	,	,	PUNCT
ijassa-372	196	23	vol.27	vol.27	NOUN
ijassa-372	196	24	,	,	PUNCT
ijassa-372	196	25	pp.1615	pp.1615	PROPN
ijassa-372	196	26	-	-	NOUN
ijassa-372	196	27	1630	1630	NUM
ijassa-372	196	28	.	.	PUNCT
ijassa-372	197	1	[	[	X
ijassa-372	197	2	16	16	NUM
ijassa-372	197	3	]	]	X
ijassa-372	197	4	c.	c.	PROPN
ijassa-372	197	5	m.	m.	PROPN
ijassa-372	197	6	bishop	bishop	PROPN
ijassa-372	197	7	.	.	PUNCT
ijassa-372	198	1	(	(	PUNCT
ijassa-372	198	2	2006	2006	NUM
ijassa-372	198	3	)	)	PUNCT
ijassa-372	198	4	,	,	PUNCT
ijassa-372	198	5	pattern	pattern	NOUN
ijassa-372	198	6	recognition	recognition	NOUN
ijassa-372	198	7	and	and	CCONJ
ijassa-372	198	8	machine	machine	NOUN
ijassa-372	198	9	learning	learning	NOUN
ijassa-372	198	10	,	,	PUNCT
ijassa-372	198	11	springer	springer	NOUN
ijassa-372	198	12	,	,	PUNCT
ijassa-372	198	13	pp.424	pp.424	PROPN
ijassa-372	198	14	-	-	X
ijassa-372	198	15	430	430	NUM
ijassa-372	198	16	.	.	PUNCT
ijassa-372	199	1	[	[	X
ijassa-372	199	2	17	17	NUM
ijassa-372	199	3	]	]	PUNCT
ijassa-372	199	4	j.	j.	PROPN
ijassa-372	199	5	c.	c.	PROPN
ijassa-372	199	6	van	van	PROPN
ijassa-372	199	7	gemert	gemert	PROPN
ijassa-372	199	8	,	,	PUNCT
ijassa-372	199	9	j.	j.	PROPN
ijassa-372	199	10	geusebroek	geusebroek	PROPN
ijassa-372	199	11	,	,	PUNCT
ijassa-372	199	12	c.	c.	PROPN
ijassa-372	199	13	j.	j.	PROPN
ijassa-372	199	14	veenman	veenman	PROPN
ijassa-372	199	15	,	,	PUNCT
ijassa-372	199	16	et	et	PROPN
ijassa-372	199	17	al	al	PROPN
ijassa-372	199	18	.	.	PUNCT
ijassa-372	199	19	(	(	PUNCT
ijassa-372	199	20	2008	2008	NUM
ijassa-372	199	21	)	)	PUNCT
ijassa-372	199	22	,	,	PUNCT
ijassa-372	199	23	“	"	PUNCT
ijassa-372	199	24	kernel	kernel	NOUN
ijassa-372	199	25	codebooks	codebook	NOUN
ijassa-372	199	26	for	for	ADP
ijassa-372	199	27	scene	scene	NOUN
ijassa-372	199	28	categorization	categorization	NOUN
ijassa-372	199	29	”	"	PUNCT
ijassa-372	199	30	,	,	PUNCT
ijassa-372	199	31	10th	10th	ADJ
ijassa-372	199	32	european	european	ADJ
ijassa-372	199	33	conference	conference	NOUN
ijassa-372	199	34	on	on	ADP
ijassa-372	199	35	computer	computer	NOUN
ijassa-372	199	36	vision	vision	NOUN
ijassa-372	199	37	,	,	PUNCT
ijassa-372	199	38	marseille	marseille	NOUN
ijassa-372	199	39	,	,	PUNCT
ijassa-372	199	40	pp.696	pp.696	VERB
ijassa-372	199	41	-	-	SYM
ijassa-372	199	42	709	709	NUM
ijassa-372	199	43	.	.	PUNCT
ijassa-372	200	1	advances	advance	NOUN
ijassa-372	200	2	in	in	ADP
ijassa-372	200	3	systems	system	NOUN
ijassa-372	200	4	science	science	NOUN
ijassa-372	200	5	and	and	CCONJ
ijassa-372	200	6	application	application	NOUN
ijassa-372	200	7	(	(	PUNCT
ijassa-372	200	8	2015	2015	NUM
ijassa-372	200	9	)	)	PUNCT
ijassa-372	200	10	vol.15	vol.15	NOUN
ijassa-372	200	11	no.3	no.3	VERB
ijassa-372	200	12	253	253	NUM
ijassa-372	201	1	[	[	SYM
ijassa-372	201	2	18	18	NUM
ijassa-372	201	3	]	]	PUNCT
ijassa-372	201	4	f.	f.	PROPN
ijassa-372	201	5	perronnin	perronnin	PROPN
ijassa-372	201	6	,	,	PUNCT
ijassa-372	201	7	j.	j.	PROPN
ijassa-372	201	8	s	s	PROPN
ijassa-372	201	9	a	a	DET
ijassa-372	201	10	nchez	nchez	NOUN
ijassa-372	201	11	and	and	CCONJ
ijassa-372	201	12	t.	t.	PROPN
ijassa-372	201	13	mensink	mensink	NOUN
ijassa-372	201	14	.	.	PUNCT
ijassa-372	202	1	(	(	PUNCT
ijassa-372	202	2	2010	2010	NUM
ijassa-372	202	3	)	)	PUNCT
ijassa-372	202	4	,	,	PUNCT
ijassa-372	202	5	“	"	PUNCT
ijassa-372	202	6	improving	improve	VERB
ijassa-372	202	7	the	the	DET
ijassa-372	202	8	fisher	fisher	PROPN
ijassa-372	202	9	kernel	kernel	NOUN
ijassa-372	202	10	for	for	ADP
ijassa-372	202	11	large	large	ADJ
ijassa-372	202	12	-	-	PUNCT
ijassa-372	202	13	scale	scale	NOUN
ijassa-372	202	14	image	image	NOUN
ijassa-372	202	15	classification	classification	NOUN
ijassa-372	202	16	”	"	PUNCT
ijassa-372	202	17	,	,	PUNCT
ijassa-372	202	18	11th	11th	ADJ
ijassa-372	202	19	european	european	ADJ
ijassa-372	202	20	conference	conference	NOUN
ijassa-372	202	21	on	on	ADP
ijassa-372	202	22	computer	computer	NOUN
ijassa-372	202	23	vision	vision	NOUN
ijassa-372	202	24	,	,	PUNCT
ijassa-372	202	25	pp.143	pp.143	NOUN
ijassa-372	202	26	-	-	PUNCT
ijassa-372	202	27	156	156	NUM
ijassa-372	202	28	.	.	PUNCT
ijassa-372	203	1	[	[	X
ijassa-372	203	2	19	19	NUM
ijassa-372	203	3	]	]	X
ijassa-372	203	4	c.	c.	PROPN
ijassa-372	203	5	h.	h.	PROPN
ijassa-372	203	6	papadimitriou	papadimitriou	PROPN
ijassa-372	203	7	and	and	CCONJ
ijassa-372	203	8	k.	k.	PROPN
ijassa-372	203	9	stieglitz	stieglitz	PROPN
ijassa-372	203	10	.	.	PUNCT
ijassa-372	204	1	(	(	PUNCT
ijassa-372	204	2	1982	1982	NUM
ijassa-372	204	3	)	)	PUNCT
ijassa-372	204	4	,	,	PUNCT
ijassa-372	204	5	“	"	PUNCT
ijassa-372	204	6	combinatorial	combinatorial	ADJ
ijassa-372	204	7	optimization	optimization	NOUN
ijassa-372	204	8	:	:	PUNCT
ijassa-372	204	9	algorithms	algorithm	NOUN
ijassa-372	204	10	and	and	CCONJ
ijassa-372	204	11	complexity	complexity	NOUN
ijassa-372	204	12	”	"	PUNCT
ijassa-372	204	13	,	,	PUNCT
ijassa-372	204	14	prentice	prentice	NOUN
ijassa-372	204	15	hall	hall	PROPN
ijassa-372	204	16	.	.	PUNCT
ijassa-372	205	1	corresponding	correspond	VERB
ijassa-372	205	2	author	author	PROPN
ijassa-372	205	3	yan	yan	PROPN
ijassa-372	205	4	yu	yu	PROPN
ijassa-372	205	5	can	can	AUX
ijassa-372	205	6	be	be	AUX
ijassa-372	205	7	contacted	contact	VERB
ijassa-372	205	8	at	at	ADP
ijassa-372	205	9	:	:	PUNCT
ijassa-372	205	10	yuyan	yuyan	PROPN
ijassa-372	205	11	wust@163.com	wust@163.com	PROPN