Academic Journal of Science and Technology 
ISSN: 2771-3032 | Vol. 6, No. 3, 2023 

 

100 

Design	of	Digital	Image	Processing	System	Based	on	
Machine	Vision	
Xiaolei Zhong1, * 

1Yunnan University Dianchi College, Kunming 650228, China 
*Corresponding author: e-mail: 1018761837@qq.com 

 

Abstract: Digital image processing technology is a technology that has developed rapidly in recent years[1]. A series of images 
are processed by computers to improve the visual effect of the image in order to achieve the desired presentation effect. Digital 
image processing technology has been continuously developed with the innovation of computers, and its application range has 
expanded from the original image field to traffic command, signal processing, medical treatment and other fields closely related 
to people's lives. With the support of many algorithms, the speed of digital image processing technology is getting faster and 
faster, which brings a lot of convenience to people's lives, and which becomes an indispensable technology in life. With the help 
of MATLAB's GUI user interface, image processing toolbox and MATLAB's comprehensive mathematical function library, this 
article designs an image processing operating platform that realizes some image processing functions after repeated trials. 

Keywords: Digital image processing, MATLAB, GUI. 
 

1. Introduction of Digital Image 
Processing and Graphical User 
Interface 

1.1. Introduction of Digital Image Processing 
Images are the basis of people's vision and the most direct 

way of understanding the outside world. "Picture" is the 
reflection of light energy from an objective object[2]. "Image" 
means that when light energy enters the eyes, it passes 
through the optic nerve and enters the brain to form image 
information. From the way the image is recorded, the image 
types can be divided into two categories: digital and analog. 

Digital image processing uses a hardware or software 
processing platform that can convert image information into 
digital information and digital information into image analog 
information to complete image preprocessing operations, and 
process it again according to needs. For example, in modern 
high-definition TV imaging, it converts the received digital 
signal into an image signal that people see with the naked eye. 
In the field of similar image recognition such as fingerprint 
recognition and face recognition, digital image processing 
technology is to perform digitized feature extraction and 
similarity comparison of the obtained image information, and 
convert the image signal into a digital signal to achieve the 
desired processing effect. 

1.2. Graphical user interface 
The user interface realizes the function of information 

interaction between the machine and the user. Through the 
user interface, the internal operation information of the 
machine is converted into an information mode that is easy 
for people to accept[3-4]. Its main function is to meet the 
needs between control and being controlled, that is, how 
people control the machine, how the machine receives 
instructions, and how to make a corresponding response and 
return to the control interface. 

The GUI image user interface of MATLAB is an image and 
graph control window that can interact with the user. It 
contains many image control objects, such as buttons, menus, 

sliders, texts, and various dialog boxes. By writing 
independent M files corresponding to these control objects 
with specific functions, the program in the M file will be 
automatically compiled when operating these control objects. 
For example, when we drag a numeric control slider, the value 
of the variable controlled by it will change accordingly.  

The characteristic of MATLAB's GUI graphical user 
interface is that users do not need to type programs in the 
MATLAB command line when implementing certain 
functions, and when designing the GUI interface, we need to 
make reasonable use of MATLAB's super numerical 
computing capabilities and image processing capabilities, and 
accurately write M files that achieve specific functions[5]. 
Therefore, the users do not need to master the programming, 
but they can still operate it. According to actual needs, you 
can customize the function buttons that need to be 
implemented on the interface to form a simple and easy-to-
operate interface to achieve the desired processing functions. 

2. Grayscale Histogram and Image 
Transformation 

2.1. Grayscale histogram 
In the image field, the way that the different saturation 

levels of black are used as the basic tone of the image is called 
grayscale[6-7]. The brightness value of the gray-scale object 
is 0% (white) to 100% (black), and the general gray-scale 
image can be obtained by a gray-scale scanner. The image 
gray bar is shown in Figure 1. 

 

 
Figure 1. Gray bar 



 

101 

Different from black-and-white images, gray-scale images 
are composed of multiple gray levels of different tones 
between white and black. The gray-scale transformations are 
as follows: 

1) For the input image f x, y  , an output image g x, y  
will be generated after the gray level transformation T, and 
the value of each pixel of g x, y  is determined by the value 
of the corresponding input pixel of f x, y , that is g x, y
T f x, y . 

2) For the original image f x, y  , the gray value 
transformation function T f x, y  , since the gray value is 
always limited (0-255), the non-geometric transformation can 
be defined as R T r : 

2.2. Grayscale histogram 
Histogram is a chart that uses different lengths of bar 

graphs to indicate the number of values on the Y axis. It has 
the advantage of being intuitive and easy to compare the size 
of the data. Gray histogram is an important processing basis 
in image processing. According to the gray data presented, the 
image can be processed such as image linear transformation 
and image nonlinear transformation. 

The histogram can represent the probability of the pixels 
corresponding to each gray level in an image appearing in the 
total pixels of the image (that is, the ratio of the pixels 
corresponding to each gray level to the total pixels in the 
entire image). Its expression is shown in (1): 

 
𝑃 𝑟 𝑘 0,1,2, … 𝐿 1          (1) 

 
In the expression (3.1), n  represents the number of pixels 

of the k-th gray level, L represents the total number of gray 
levels, N represents the total number of pixels in the image, 
and P r   represents the probability that the pixel 
corresponding to each gray level in an image appears in the 
total pixels of the image, r  represents the kth gray level. 

3. Histogram Equalization 
Histogram equalization is an image processing method that 

changes the gray distribution of the original image, which is 
achieved through methods such as contrast value 
adjustment[8]. Histogram equalization increases the value 
range of pixels to obtain a grayscale image with uniform 
probability density. Its processing method is to modify the 
histogram through cumulative distribution function 
transformation, which is widely used in image enhancement 
processing. Assuming that the original gray value of a certain 
pixel is R, the processed gray value is S, and its gray scale 
transformation function T R  is shown in equation 2: 

 
S T R ∑ 𝑃 𝑅 ∑          (2) 

 
0 𝑅 𝐿; 𝑘 0,1 … 
In formula 3.2, it represents the probability of the gray 

value of the j-th level, n   represents the total number of 
pixels of the j-level gray in the image, 𝐿 represents the total 
number of gray levels, and 𝑛 represents the total number of 

pixels in the image. 
Histogram equalization is widely used in image gray-scale 

histogram processing. The histogram equalization can be 
implemented as follows: 

1) Perform histogram statistics on the images that need 
histogram equalization processing, and find 

 
𝑃 (𝑟 )=                 (3) 

 
2) Perform further cumulative distribution function 

transformation processing on the processed gray histogram to 
obtain a new gray level; 

 
𝑆 𝑇 𝑅 ∑ 𝑃 𝑅          (4) 

 
3) Integrate the gray value approximately equal to one, and 

replace the old gray with the new gray. 

4. Linear Transformation of Images 
This article adopts the transformation method of point 

arithmetic for linear transformation. Point operation is 
actually the process of calculating the gray value of each pixel 
of the image according to a certain mapping relationship to 
obtain a new image. That is, for the pixel gray value f x, y  
of the original image, the pixel value g x, y   of the output 
image will be generated by the gray level transformation T, 
and each pixel value of g x, y  is determined by the value of 
the corresponding input pixel of f x, y  , thus achieving the 
point-to-point arithmetic processing on the image, this kind 
of arithmetic processing is called point arithmetic. 

r  represents the gray level index of the input image, s 
represents the gray level index of the output image, T 
represents the functional mapping relationship between the 
image input value and the image output value, as shown in 
Equation 5and Equation 6: 

 
s T r                    (5) 

 
g x, y T f x, y               (6) 

 
The grayscale transformation is used to enhance or lighten 

the grayscale of each pixel of the original image, so that the 
grayscale dynamic range of the image is changed. The 
contrast of the image is enhanced to meet the requirements of 
improving the visual effect of the image. 

According to the different requirements of the image gray-
scale linear transformation processing, the slope 𝑘  of the 
transformation function can be changed to meet the 
processing requirements. Supposing the original image gray 
value f x, y ∈ 𝑎, 𝑏  , the image gray value g x, y ∈ 𝑐, 𝑑  
after linear transformation, the transformation relationship (7) 
can be obtained: 

 
g x, y c k f x, y 𝑎            (7) 

 

𝑘                   (8) 

 
 



 

102 

 
Figure 2. Gray linear transformation relationship 

 
According to the value of sum, it can be roughly divided 

into the following situations: 
1) Dynamic range expansion: if 𝑎, 𝑏 ∈ 𝑐, 𝑑 , that is k>1, 

then the gray value range of the image will be expanded, and 
the current situation of underexposure of the image will be 
improved, and the full use of the image display instrument 
will be improved; 

2) Change the value range: If k 1, that is 𝑑 𝑐 𝑏
𝑎 , then the gray dynamic range after transformation is the 
same as the original, but the gray value range will change with 
the changes of 𝑎 and 𝑐; 

3) Reduce the dynamic range: If 𝑐, 𝑑 ∈ 𝑎, 𝑏 , that is 0
k 1 , the dynamic range of the processed image will 
decrease accordingly; 

4) Image inversion: If k 0, for 𝑏> 𝑎, the gray value of 
the image will be inverted after the transformation process, 
that is, the original image gray value will be reversed 180 
degrees, which is simply to change the image from "black" to 
"white", or from "white" to "black". At k 1 , g x, y   is 
the inverse of f x, y . 

4.1. Piecewise linear transformation of images 
Image segmentation divides the image into many parts. The 

overall histogram of an image can be obtained by 
superimposing the histograms of each part. By highlighting 
the part of the gray interval, that is, linearly transforming the 
gray value of the interval, it can highlight the part that people 
are interested in, image processing methods like this are 
called image piecewise linear transformation methods. 

In the piecewise linear transformation, in order to highlight 
the details of the interesting part in the image, the gray level 
of the output is often enhanced, and the uninteresting part will 
be suppressed to a certain extent, and the gray level will also 
be weaken. Supposing the interval of interest is the pixel gray 
value of the input image, and the pixel value of the output 
image is generated after piecewise linear change. The 
following formula can be obtained: 

 

g x, y

⎩
⎪
⎨

⎪
⎧ f x, y ;               0 f x, y  𝑎  

f x, y 𝑎 𝑐;     𝑎 f x, y 𝑏   

f x, y 𝑏 𝑑;     0 f x, y  𝑎  

 (9) 

 

 

 
Figure 3. The relationship of gray-scale piecewise linear transformation 

 

4.2. Non-linear transformation of images 
When imaging an image, it may be interfered by some 

uncontrollable factors in the environment, such as noise or 
other signals that will randomly affect the output image, 
resulting in non-linear distortion of the output image. At this 
time, it is necessary to perform a certain amount of non-
control on the output image to meet image imaging 
requirements. In this paper, the exponential transformation 

method is used as the research method of image nonlinear 
transformation. 

1) Exponential transformation 
We adopt the method of expanding the high-value gray-

scale area of the image and compressing the low-value gray-
scale area, that is, the exponential transformation processing 
method. Its expression is shown in formula (10): 

 
g x, y 𝑏 , 1           (10) 



 

103 

The exponential transformation curve of the image is 
shown in Figure 4. 

 

 
Figure 4. Exponential transformation curve 

 

4.3. Image smoothing 
The purpose of image smoothing is to improve image 

quality through specific algorithms in accordance with image 
processing requirements, thereby it can enhance people's 
visual effects. Image smoothing processing technology is a 
commonly used processing method in digital image 
processing technology. Its main function is to reduce the high-
frequency components in the image, reduce the sharpness of 
the image and remove some noise. The high-frequency 
components in the image represent the edge contours and high 
gray value areas of the image. Through image smoothing, the 
high-frequency components in the image can be well 
suppressed, so that the overall grayscale distribution is 
uniform, and the image looks more uniform. This article uses 
the domain average and median filter method to smooth the 
image. 

4.4. Image sharpening 
Image sharpening and image smoothing are two different 

image processing methods. Image sharpening is to make the 
edge of the image clear, make the image outline more 
prominent, and achieve a presentation form that is convenient 
for observing the image outline. In essence, image sharpening 
is to strengthen the high-frequency part of the image, and then 
the image noise will be strengthened. After research and 
processing, the edge of the image is detected by the method 
of edge detection first, and then based on the detection result, 
the high-frequency components of the image are enhanced. 

In the image sharpening method, image processing is 
mainly to do edge enhancement processing on the prominent 
lines or isolated points in the image, and it also enhances the 
image noise signal, that is, the purpose of sharpening is to 
make the points with large grayscale pixels in the contour of 
the image to make their grayscale values larger, and the pixels 
with small grayscale values become smaller, while the pixel 
point outside the outline is not affected by sharpening. 

4.5. Image sharpening algorithm 
In order to make the edges of the image clear and make the 

outline of the image more prominent, isotropic gradient 
algorithms are often used. The basic expression is shown in 
equation (11): 

 

        (11) 

 
In the above formula, x, y  is the image before processing, 

and 𝑥 , 𝑦  is the image after processing. The gradient value 
of the image x, y  point is shown in equation (12): 

g f kz∇ 𝑓              (12) 

 
In the formula (4.5), k represents the coefficient related to 

the diffusion effect. Through the improvement of the gradient 
value algorithm, the edge of the object can be effectively 
made prominent, and ∇ 𝑓 represents the Laplacian operator 
that performs second-order differential processing on the 
image 𝑓. ∇  can be represented by a template, such as (13): 

 

H
2 4 2
4 20 4
2 4 2

            (13) 

 
In practical applications, the value of each coefficient in the 

template H can be changed. In formula (4.5), the choice of 
value is also very important. If the value is too large, the edge 
of the image will be too prominent and affect the visual effect. 
If the value is too small, the sharpening effect will be 
insignificant. After repeated trials and verification, the value 
of kz is between 3 and 8, and the treatment effect is the best. 
The effect after sharpening is shown in Figure 5 and Figure 6: 

 

 
Figure 5. Original image 

 

 
Figure 6. The sharpened image 

 
Compared with the original image, after the image is 

sharpened, its edge contours are easier to observe, which 
provides people with a good basis for image features such as 
image processing operations such as photo restoration. 



 

104 

5. Conclusion 
This article uses MATLAB's powerful scientific computing 

capabilities to compile the digital image processing program, 
and finally present the image processing results through the 
MATLAB graphical user interface. In this way, different 
processing functions can have independent M files. ,It is more 
concise and clear when writing and debugging programs. 

References 
[1] Zhang Guangcai, Wan Shoupeng.Development and Design of 

MATLAB GUI Digital Image Processing System [J]. Software, 
2019, 40(11): 123-127. 

[2] Han Zhaoyang, Weng Dongdong. An Optical Tracking System 
Based on Simple Marker Coding[J]. Acta Graphics: 1-16. 

[3] Zhang Lili, Bao Xirong, Wang Tong, etc. Research on 
Integrated Circuit Welding Quality Monitoring System Based 
on Image Processing Technology [J]. Electronic Devices, 
2023,46(01):57-61. 

[4] Xu Chunhe, Sun Peigang, Zhang Quanyu, etc. Design of 
Digital Image Processing System Based on Machine Vision [J]. 
Popular Standardization, 2022, No.379(19): 128-130. 

[5] He Zhichao, Fan Yuhua, etc. Research on Image Processing 
Optimization Technology of Railway Passenger Station Video 
Fusion Intelligent Monitoring System Based on Deep Learning 
[J].Railway Computer Application, 2022,31(07):81-86. 

[6] Yan Zongyi, Ren Dejun, etc. Design of Deep Learning 
Platform for Image Processing Based on B/S Architecture [J]. 
Modern Electronic Technology, 2022, 45(16): 60-66. 

[7] Yin Feihao, Shang Jianhua. Research on Image Processing 
Algorithm IP Verification System Based on ZYNQ Platform 
[J]. Computer Technology and Development, 2022,32(10):41-
45. 

[8] Sun Pengwei, Wang Jun, Wang Shujun, etc. Design of Image 
Processing System Based on MATLAB GUI [J]. Computer 
Technology and Development, 2022,32(04):215-220.