Adv Syst Sci Appl 2019; 03; 38-51 Published online at http://ijassa.ipu.ru/index.php/ijassa/article/view/724 Improved Steganography Techniques For Different Types Of Secret Data Rasha Thabit* Computer Techniques Engineering Department, Al-Rasheed University College, Hay Al- Hussain, Baghdad, Iraq E-mail: rashathabit@yahoo.com Received April 9, 2019; Revised September 18, 2019; Published October 1, 2019 Abstract: Steganography is an interesting aspect of data hiding art where the secret data such as (binary bits, logo, image, audio, and others) are embedded in a cover media without raising any suspicion about the existence of that data. Over the years, different steganography techniques for digital images have been presented to be applied in various applications such as security and privacy control. Some of these techniques are based on transforming the image from the spatial domain to the transform domain as a pre-processing step before embedding the secret data. This paper presents two improved steganography techniques in the transform domain based on Slantlet transform (SLT) and different data embedding methods. In the first proposed technique, a secret image can be embedded in a cover image by calculating the difference blocks of SLT coefficients and modifying the coefficients according to the root mean square error values. In the second proposed technique, a sequence of binary bits can be embedded in a cover image by modifying the carrier SLT subbands according to the secret data bits and the mean values of SLT coefficients for image’s blocks. The experimental results proved the efficiency of the proposed techniques in comparison with the previous transform-based steganography techniques. Keywords: Data hiding, Steganography, Slantlet transform, Security, Sharing secret data. 1. INTRODUCTION Data hiding techniques are widely used in digital communication for different purposes such as security and privacy control [1]. The steganography [2-5] and watermarking [6,7] techniques are the two main parts of data hiding science. These techniques are based on hiding secret data in the cover media either by directly changing the contents of the cover media (i.e., embedding in spatial domain) or by transforming the cover media and embedding the secret data in the resultant coefficients after transformation (i.e., embedding in the transform domain). In the transform-based data hiding techniques, different transforms have been applied as a pre-processing step before embedding the secret data such as discrete cosine transform (DCT) [8, 9], discrete wavelet transform (DWT) [5, 10, 11], and an orthogonal DWT called Slantlet transform (SLT) [4,12,13]. On the other hand, the implementation of the steganography algorithm is affected by the type of secret data that is intended to be embedded in the cover image. In [14], a steganography algorithm has been presented to hide a secret text in a cover image by embedding each character from the text in a window of pixels of size 5×5 using a rule of non-multiples of 5. In [15], a secret text file is compressed and embedded in the pixels of the Red, Green, and Blue (RGB) channels of the color image. In [16], a text message is converted to a sequence of binary bits and embedded in the least significant bits (LSB) of the pixels in the cover image. In [4], another steganography technique has been introduced to embed binary bits in a cover image using Slantlet transform (SLT) [17] and two data *Corresponding author: rashathabit@yahoo.com mailto:rashathabit@yahoo.com THE EFFECT OF TWO MAJOR FACTORS ON THE PERFORMANCE OF STEGANOGRAPHY 39 Copyright ©2019 ASSA. Adv. in Systems Science and Appl. (2019) embedding methods (i.e., LSB Substitution method and the threshold embedding method). The performance of the steganography technique has been improved in comparison with the DWT-based scheme, however, the visual quality of the stego images still needs more improvement. Other steganography techniques have been presented to hide a secret image in a cover image. In [18], a secret image is embedded in the LSBs of a cover image which could be a color or grayscale image. The exclusive-or (XOR) of a block of pixels is used to determine the smoothness of the neighborhood in order to obtain the number of bits that can be embedded in the cover image. In [5], a DWT-based steganography technique has been presented to hide a secret image in a cover image. The technique converts the cover image and the secret image using DWT and the resultant coefficients are divided into 4-subbands named as approximation, horizontal, vertical, and the diagonal coefficients. Each subband is divided into non-overlapping blocks then the approximation coefficients of the secret image are compared with the approximation coefficients of the cover image and the root mean squared error (RMSE) is calculated to find the locations of the best similar blocks. The differences between these coefficients are calculated and saved as error blocks then the RMSE is used to find the locations of the best matched horizontal coefficients blocks with the error blocks, thereafter, the error blocks are embedded in the horizontal coefficients blocks. The work in this research paper is directed towards the steganography techniques in the transform domain and its aims can be summarized as follows:  Introducing an improved steganography technique to hide secret image in a cover image.  Introducing another improved steganography technique to hide a sequence of binary bits in a cover image. To meet the first aim of this work, the transform type (i.e., DWT) is changed to (SLT) in a previous steganography technique [5] while keeping the same data embedding method then the data embedding method is modified to improve the performance. To meet the second aim of this work, an improved steganography technique is introduced to hide a sequence of binary bits in a cover image by modifying the carrier SLT subbands according to the secret data bits and the mean values of SLT coefficients for the cover image’s blocks. The rest of the paper is organized as follows: section 2 explains the implemented algorithms, section 3 illustrates the experiments and their discussion, and section 4 contains the conclusion. 2. THE IMPLEMENTED TECHNIQUES The following subsections presents the steps of the proposed algorithms in this paper. The first algorithm (SLT-EM1) is implemented to hide a secret image in a cover image using SLT and an embedding algorithm from [5]. Then the same algorithm has been modified to improve the performance in terms of visual quality and the modified algorithm is named as SLT-MEM1. The second steganography algorithm has been implemented to embed a sequence of binary bits in a cover image and it is named as SLT-EM2. The abbreviations of the proposed algorithms can be explained as follows:  SLT-EM1: refers to Slantlet Transform and Embedding Method 1.  SLT-MEM1: refers to Slantlet Transform and Modified Embedding Method 1.  SLT-EM2: refers to Slantlet Transform and Embedding Method 2. 2.1. SLT-EM1 The algorithms of SLT-EM1 for the secret image embedding and extraction processes are summarized in Fig. 1 and Fig. 2, respectively. The detailed steps are explained in the following subsections. 40 R. THABIT Copyright ©2019 ASSA Adv. in Systems Science and Appl. (2019) 2.1.1. SLT-EM1 data embedding procedure Step 1: Read the original cover image ( oI ) and the secret image ( sI ). Step 2: Apply the 2D-SLT to oI and divide the resulting coefficients into four subbands ( CDandCVCHCA ,,, ). Fig. 1. SLT-EM1 data embedding procedure Step 3: Apply the 2D-SLT to sI and divide the resulting coefficients into four subbands ( SDandSVSHSA ,,, ). Step 4: Divide SAandCHCA ,, subbands into non-overlapping blocks of size 4×4. The subbands’ blocks can be represented as follows: }1,{ }1,{ }1,{ sk i i nkBSSA niBHCH niBACA    Where iBA is the ith block in CA , iBH is the ith block in CH , and kBS is the kth block in SA , n is the total number of 4×4 blocks in CA and in CH , sn is the total number of 4×4 blocks in SA . Step 5: Calculate the RMSE to find the best matched block iBA with kBS , save the locations i of these best matched blocks in a secret key 1K . Step 6: Calculate the difference (i.e., error block) kEB between iBA and kBS as follows: kik BSBAEB  …(1) Step 7: For each block in kEB , use the RMSE to find the best matched block iBH ; save the locations i of these best matched blocks in a secret key 2K . Then replace iBH with kEB . Step 8: Continue to embed all the difference blocks in CH then replace the original CH subband with the new CH subband. Step 9: To obtain the stego image G apply the inverse 2D-SLT to ( CDandCVCHnewCA ,,, ). THE EFFECT OF TWO MAJOR FACTORS ON THE PERFORMANCE OF STEGANOGRAPHY 41 Copyright ©2019 ASSA. Adv. in Systems Science and Appl. (2019) The previous algorithm in [5] searches for the locations of the best matched blocks to obtain (K1 and K2) but some of the locations may be repeated more than once and this will cause losing some of the error blocks (because they will be embedded in the same location). To avoid this problem, the embedding algorithm in this paper has been modified to check the locations that has been obtained and if the same location appeared again then the location of the next block with the minimum RMSE must be found. 2.1.2. SLT-EM1 data extraction procedure Fig. 2. SLT-EM1 data extraction procedure Step 1: Apply the 2D-SLT to the stego image G and divide the resulting coefficients into four subbands ( GCDandGCVGCHGCA ,,, ). Step 2: Divide GCHandGCA, into non-overlapping blocks of size 4×4. The blocks can be represented as follows: }1,{ }1,{ niBHGCH niBAGCA i i   Step 3: Use secret key 1K to find the iBA and use the secret key 2K to extract the error blocks kEB from iBH . The secret image block kBS can be recovered as follows: kik EBBABS  …(2) Step 4: Repeat (step 3) to extract all the secret blocks. Rearrange the blocks to obtain SA. Step 5: Set SDandSVSH ,, as zeros then apply inverse 2D-SLT to ( SDandSVSHSAExtracted ,,, ) to obtain the secret image. To embed the secret image in CDorCV, subbands, the same procedures can be used but the CH subband must be replaced by the required subband. 42 R. THABIT Copyright ©2019 ASSA Adv. in Systems Science and Appl. (2019) In order to improve the visual quality of the stego image, the embedding algorithm is modified and the technique is named as SLT-MEM1. In the modified algorithm the error blocks are divided by a variable α and the procedure of the embedding process (section 2.1.1) is the same but in step 6 the following equation will be used: )( 1 kik BSBAEB   …(3) In the extraction process (section 2.1.2) the same procedure repeated but in step 3 the following equation will be used: kik EBBABS   …(4) 2.2. The proposed steganography technique (SLT-EM2) The previous study of SLT-EM1 and SLT-MEM2 proved that SLT performs better in comparison with DWT, therefore it will be adopted in the proposed steganography technique (SLT-EM2) in which a new data embedding method is suggested in order to improve the performance. The proposed embedding process depends on modifying the carrier SLT subband according to the secret data bit and the mean value of SLT coefficients in a block. The algorithms of SLT-EM2 for the secret data embedding and extraction processes are summarized in the block diagrams shown in Fig. 3 and Fig. 4, respectively. The steps of the algorithms are detailed in the following subsections. 2.2.1. SLT-EM2 data embedding procedure Step 1: Apply 2D-SLT to the cover image ( mI ) and divide the coefficients into four subbands ( CDandCVCHCA ,,, ). Step 2: Divide each subband ( CDandCVCH ,, ) into non-overlapping blocks of size ( ss BB  ) as follows: }1,{ }1,{ }1,{ nciBCDCD nciBCVCV nciBCHCH i i i    Where iBCH is the ith block in CH, iBCV is the ith block in CV, iBCD is the ith block in CD, nc is the total number of ( ss BB  ) blocks in each subband. Step 3: Calculate the mean value of the coefficients in each block as follows:     s sB x B y yx s block C B M 1 1 ,2 1 …(5) }1,{ }1,{ }1,{ nciMCDCDMean nciMCVCVMean nciMCHCHMean i i i    THE EFFECT OF TWO MAJOR FACTORS ON THE PERFORMANCE OF STEGANOGRAPHY 43 Copyright ©2019 ASSA. Adv. in Systems Science and Appl. (2019) Where blockM is the mean value for the block of size ss BB  , yxC , is the SLT coefficient in the coordinates ( yx, ), iMCH is the mean value of the ith block in CH , iMCV is the mean value of the ith block in CV , iMCH is the mean value of the ith block in CD . Step 4: The rules of the proposed data embedding method can be explained as follows:  Each bit from a binary sequence of data ( kw ) {where k=1, 2, …, Length ( w )} can be embedded using two successive blocks. The main idea of the proposed data embedding process is that, when kw is ‘1’ then the mean value of the first block should be more than the mean value of the second block, when kw is ‘0’ the reverse is true. The embedding process in CH will be explained and the same idea can be applied for embedding data in CV and CD .  To embed a binary bit kw in two successive blocks the mean values of these blocks will be used to control the embedding process as follows: Let 121 ,  ii MCHxMCHx If 211 xxandwk  , then exchange the blocks 1ii BCHandBCH Else if 211 xxandwk  , then do nothing; If 120 xxandwk  , then exchange the blocks 1ii BCHandBCH Else if 120 xxandwk  , then do nothing; This process must be repeated until embedding all data bits. Step 5: Rearrange the blocks to construct the modified subbands. Step 6: Apply inverse 2D-SLT to obtain the stego image. Fig. 3. SLT-EM2 data embedding procedure 44 R. THABIT Copyright ©2019 ASSA Adv. in Systems Science and Appl. (2019) 2.2.2. SLT-EM2 data extraction procedure Step 1: Apply the SLT to the stego image (G ) and divide the coefficients into four subbands ( GCDandGCVGCHGCA ,,, ). Step 2: Divide each subband ( CDandCVCH ,, ) into non-overlapping blocks of size ( ss BB  ) as explained in (section 2.2.1, step 2). Step 3: Calculate the mean value of the coefficients for each block in subband as explained in (section 2.2.1, step 3). Step 4: The length of the data is required at the receiver side where the data bits can be extracted as follows: )(...,,2,1, ,0 ,1 1 1 wlengthkfor MCHMCHif MCHMCHif w ii ii k          Fig. 4. SLT-EM2 data extraction procedure 3. THE EXPERIMENTAL RESULTS AND DISCUSSION This section presents the experiments that have been conducted to evaluate the implemented algorithms: SLT-EM1, SLT-MEM1 and SLT-EM2. The first subsection contains the performance evaluation of the implemented algorithms (SLT-EM1 and SLT-MEM1) in comparison with the DWT-based technique from [5]. The second subsection contains the performance evaluation of the proposed technique (SLT-EM2) in comparison with the previous SLT-based technique from [4]. In [4,5], the metric that have been used to evaluate the visual quality of the stego image is the Peak Signal-to-Noise Ratio (PSNR), therefore, the same metric is adopted here to make the comparison valid. 3.1. Experimental Results of SLT-EM1 and SLT-MEM1 For comparison purpose, the same test images that have been used in [5] are utilized in the experiments of this section. The cover images (Peppers, Goldhill, Cameraman, and Barbara) are shown in Fig. 5 where each image is of size (256×256) pixels. Fig. 6 contains the secret images (Airplane and Bird) with size (128×128) pixels. THE EFFECT OF TWO MAJOR FACTORS ON THE PERFORMANCE OF STEGANOGRAPHY 45 Copyright ©2019 ASSA. Adv. in Systems Science and Appl. (2019) (a) (b) (c) (d) Fig. 5. Cover images: (a) Peppers, (b) Goldhill, (c) Cameraman, and (d) Barbara. (a) (b) Fig. 6. Secret images: (a) Airplane, and (b) Bird. The secret images are embedded in CH subband of the cover images using (SLT-EM1 and SLT-MEM1); α in SLT-MEM1 is set to 2 (as an example). Fig. 7 presents the PSNR results of the stego images for the two secret images (i.e., Airplane and Bird). The results proved that, replacing the DWT by SLT improved the visual quality results of the stego images, however, further improvement has been obtained using SLT-MEM1 in which the embedding method has been modified. The PSNR between the original secret image and the extracted secret image has been calculated to evaluate the quality of the extracted secret image and the results are shown in Table 1. As illustrated in the results, the extracted secret images using SLT-EM1 and SLT- MEM1 obtained better visual quality results in comparison with the DWT-based scheme [5]. The difference between SLT-EM1 and SLT-MEM1 is only in the embedding process, therefore, the same secret information is embedded and extracted and this is proved in the results of Table 1. (a) (b) Fig. 7. PSNR(dB) results for the stego images when the secret image: (a) Airplane, and (b) Bird. 46 R. THABIT Copyright ©2019 ASSA Adv. in Systems Science and Appl. (2019) Table 1 The PSNR (dB) of the extracted secret images from CH subband Cover images (256×256) Secret images (128×128) Airplane Bird DWT [5] SLT-EM1 SLT-MEM1 DWT [5] SLT-EM1 SLT-MEM1 Peppers 22.3544 25.2799 25.2796 28.8265 32.0501 32.0500 Goldhill 22.3544 25.2793 25.2793 28.8265 32.0506 32.0502 Cameraman 22.3544 25.2794 25.2795 28.8265 32.0505 32.0502 Barbara 22.3544 25.2791 25.2791 28.8265 32.0500 32.0500 The effect of embedding the secret images in different subbands ( CDandCVCH ,, ) of the cover images has been studied and the results of SLT-EM1 and SLT-MEM1 are shown in Table 2 and Table 3, respectively. As illustrated in Table 2 and Table 3, the obtained results for CH and CV subbands are convergent and both are better than the results of CD subband; on the other hand, the SLT-MEM1 technique obtained better results in comparison with SLT-EM1 technique. Table 2 The PSNR(dB) results for the stego images at different subbands using SLT-EM1 Cover images (256×256) Secret images (128×128) Airplane Bird CH subband CV subband CD subband CH subband CV subband CD subband Peppers 28.2451 28.4586 27.8485 29.6938 29.6936 29.1933 Goldhill 28.3697 28.2927 27.7174 29.0134 29.2456 28.4965 Cameraman 28.6842 28.8816 27.7867 29.8682 30.0028 29.2273 Barbara 28.1230 28.2932 27.7850 29.4041 29.6504 29.0477 Table 3 The PSNR(dB) results for the stego images at different subbands using SLT-MEM1 Cover images (256×256) Secret images (128×128) Airplane Bird CH subband CV subband CD subband CH subband CV subband CD subband Peppers 32.6267 32.7314 32.3668 34.1096 33.9971 33.2923 Goldhill 32.6697 32.5951 32.0575 33.0514 33.2603 32.4745 Cameraman 33.1888 33.3216 32.5160 33.8336 33.8251 33.3362 Barbara 32.5431 32.6211 32.2960 33.6547 33.7584 33.4449 3.2. Experimental Results of SLT-EM2 In order to prove the efficiency of the proposed data embedding method, the experiments have been conducted to compare the performance of the proposed SLT-based steganography scheme with a previous SLT-based steganography scheme from [4] which used two different embedding methods (i.e., LSB and threshold). For comparison purposes, the experiments have been conducted for the same test image (i.e., grayscale Lena image) that has been used in [4]. In the proposed steganography scheme (SLT-EM2), different block sizes can be used and the data can be embedded in different subbands, therefore, the experiments have been conducted to evaluate the visual quality for the three subbands ( CDandCVCH ,, ) at different block sizes ( ss BB  ). Fig. 8 and Fig. 9 present the experimental results and comparison with the LSB embedding method from [4] for the payload (36 bits) at block size (2×2) and block size (4×4), respectively. The results proved that the stego image using the proposed SLT-EM2 technique obtained better visual quality results in comparison with the LSB embedding method. The highest PSNR values are obtained when the secret data is embedded in CD subband. On the other hand, the THE EFFECT OF TWO MAJOR FACTORS ON THE PERFORMANCE OF STEGANOGRAPHY 47 Copyright ©2019 ASSA. Adv. in Systems Science and Appl. (2019) results proved that in SLT-EM2, the lower the block size the better the visual quality results because less changes will be conducted in the stego image. Fig. 8. Visual quality comparison of SLT-EM2 with LSB method at block size (2×2). Fig. 9. Visual quality comparison of SLT-EM2 with LSB method at block size (4×4). To compare the performance of SLT-EM2 and the threshold embedding method from [4], the same previous experiments have been conducted but the payload is changed to (44 bits). Fig. 10 and Fig. 11 present the experimental results which proved that the proposed SLT- EM2 technique obtained better visual quality results in comparison with the threshold embedding method. 48 R. THABIT Copyright ©2019 ASSA Adv. in Systems Science and Appl. (2019) Fig. 10. Visual quality comparison of SLT-EM2 with threshold method at block size (2×2). Fig. 11. Visual quality comparison of SLT-EM2 with threshold method at block size (4×4). The visual quality using SLT-EM2 has been tested at higher payloads and the comparison results are illustrated in Table 4. The results proved that for higher payloads the stego image using SLT-EM2 has better visual quality in comparison with LSB and threshold embedding methods. Table 4 Visual quality comparison at higher payloads Payload (bits) Threshold Embedding [4] LSB Embedding [4] SLT-EM2 1000 ― ― 39.8666 1673 24.6087 25.8256 36.7899 3727 24.6042 25.8249 31.7497 4000 ― ― 31.1905 Another experiment has been conducted to evaluate the error level in the extracted binary bits for the proposed algorithm SLT-EM2 by calculating the normalized correlation (NC) between the original bits and the extracted bits. Table 5 presents the NC results for binary sequences of length (1000 and 1500 bits) which are embedded in the test images that have been used in the previous experiments (i.e., Lena, Peppers, Goldhill, Cameraman, and Barbara) at block sizes (2×2 and 4×4). The results proved that there is a small amount of error in the extracted bits which is resulted from the rounding operation due to the embedding in the transform domain. THE EFFECT OF TWO MAJOR FACTORS ON THE PERFORMANCE OF STEGANOGRAPHY 49 Copyright ©2019 ASSA. Adv. in Systems Science and Appl. (2019) Table 5 Normalized Correlation (NC) results Test image Binary Sequence (1000 bits) Binary Sequence (1500 bits) block size (2×2) block size (4×4) block size (2×2) block size (4×4) Lena 0.8960 0.9580 0.9150 0.9680 Peppers 0.9520 0.9530 0.9520 0.9467 Goldhill 0.9620 0.9860 0.9707 0.9575 Cameraman 0.8940 0.9201 0.8933 0.9094 Barbara 0.9500 0.9600 0.9402 0.9494 4. CONCLUSION This paper presents two improved steganography techniques based on SLT and different embedding methods. The first proposed technique embeds secret image in a cover image while the second proposed technique embeds binary data in a cover image. The work in this paper proved that the change of the transform type while using the same data embedding method can improve the visual quality of the stego images, on the other hand, the change of the data embedding method while applying the same transform type can also improve the performance of the steganography technique. The experimental results of the first technique proved that the SLT-based steganography techniques perform better in comparison with the DWT-based techniques. In addition, the modification of the embedding algorithm causes further improvement in the results. The experimental results of the second technique proved the efficiency of the proposed technique in comparison with a previous SLT-based steganography technique. From the work presented in this paper, one can conclude that the researcher who intend to implement a new steganography technique in the transform domain should investigate the performance of his / her technique using different transforms to choose the best transform that can match the embedding method to obtain the optimum results. REFERENCES 1. Hassan, N. A., & Hijazi, R. (2016). Data hiding techniques in windows OS: A practical approach to investigation and defense, Syngress, 1st edition, 324 pages. https://doi.org/10.1016/C2015-0-02014-7 2. Nosrati, M., Karimi, R., & Hariri, M. (2011). An introduction to steganography methods, World Applied Programming, 1(3), 191-195. https://www.semanticscholar.org/paper/An-introduction-to-steganography-methods- Nosrati/23311184a7b078945f519e8bf89c719fed7b1f81 3. Sumathi, C. P., Santanam, T., & Umamaheswari, G. (2013). A Study of Various Steganographic Techniques Used for Information Hiding, International Journal of Computer Science & Engineering Survey (IJCSES), 4 (6), 9-25. https://arxiv.org/ftp/arxiv/papers/1401/1401.5561.pdf 4. Kumar, S., & Muttoo, S. (2009). Distortionless Data Hiding based on Slantlet Transform, International Conference on Multimedia Information Networking and Security, Hubei, China, vol. 1, 48-52, 18-20 November. https://doi.org/10.1109/MINES.2009.263 5. Kumar, V., & Kumar, D. (2010). Performance Evaluation of DWT Based Image Steganography, IEEE 2nd International Advance Computing Conference (IACC), 223-228, Patiala. https://doi.org/10.1109/IADCC.2010.5423005 https://www.semanticscholar.org/paper/An-introduction-to-steganography-methods-Nosrati/23311184a7b078945f519e8bf89c719fed7b1f81 https://www.semanticscholar.org/paper/An-introduction-to-steganography-methods-Nosrati/23311184a7b078945f519e8bf89c719fed7b1f81 https://arxiv.org/ftp/arxiv/papers/1401/1401.5561.pdf https://doi.org/10.1109/MINES.2009.263 https://doi.org/10.1109/IADCC.2010.5423005 50 R. THABIT Copyright ©2019 ASSA Adv. in Systems Science and Appl. (2019) 6. Tao, H., Chongmin, L. J., Zain, M. A., & Abdalla, N. (2014). Robust Image Watermarking Theories and Techniques: A Review, Journal of Applied Research and Technology, 12(1), 122-138. https://doi.org/10.1016/S1665-6423(14)71612-8 7. Dixit, A., & D., Rahul. (2017). A Review on Digital Image Watermarking Techniques, I. J. Image, Graphics and Signal Processing, vol. 4, 56-66. https://doi.org/10.5815/ijigsp.2017.04.07 8. Amer, I., Hishmat, P., Badawy, W., Jullien, G. (2010). Comparisons and Analysis of DCT-based Image Watermarking Algorithms, K. Elleithy (ed.), Advanced Techniques in Computing Sciences and Software Engineering, Springer Science + Business Media B. V., 55-58. https://doi.org/10.1007/978-90-481-3660-5_10 9. Xu, Z.J., Wang, Z.Z., & Lu, Q. (2011). Research on Image Watermarking Algorithm Based on DCT, Procedia Environmental Sciences,10(B), 1129-1135. https://doi.org/10.1016/j.proenv.2011.09.180 10. Kamesh, S., Durga Devi, K., & Raviteja, S.N.V.P. (2017). DWT BASED DATA HIDING USING VIDEO STEGANOGRAPHY, International Journal of Engineering Science & Research Technology, 6(4), 361-367. https://doi.org/10.5281/zenodo.556248 11. AlZain, M. A., & Al-Amri, J. F. (2018). Application of Data Steganographic Method in Video Sequences Using Histogram Shifting in the Discrete Wavelet Transform, International Journal of Applied Engineering Research, 13(8), 6380-6387. https://www.ripublication.com/ijaer18/ijaerv13n8_106.pdf 12. Mohammed, R. T., & Khoo, B. E. (2013). Robust reversible watermarking scheme based on wavelet-like transform, 2013 IEEE International Conference on Signal and Image Processing Applications, Melaka, 354-359. https://doi.org/10.1109/ICSIPA.2013.6708032 13. Thabit, R., & Khoo, B. E. (2014). Capacity improved robust lossless image watermarking, IET Image Processing, 8(11), 662-670, 2014. https://doi.org/10.1049/iet-ipr.2013.0862 14. Jassim, F. A. (2013). A Novel Steganography Algorithm for Hiding Text in Image using Five Modulus Method, International Journal of Computer Applications, 72(17), 39-44. https://arxiv.org/ftp/arxiv/papers/1307/1307.0642.pdf 15. Sharma, V., & Kumar, S. (2013). A New Approach to Hide Text in Images Using Steganography, International Journal of Advanced Research in Computer Science and Software Engineering, 3(4), 701-708. https://pdfs.semanticscholar.org/1243/57f41f1b3a87d7380763d2a78041786a35be.pdf 16. Kaul, N., Bajaj, N. (2013). Audio in Image Steganography based on Wavelet Transform, International Journal of Computer Applications, 79(3), 7-10. https://pdfs.semanticscholar.org/4d0f/50b782f10b01e5e963088d07d25037096263.pd f 17. Selesnick, I. W. (1999). The Slantlet Transform, IEEE Transactions on Signal Processing, 47(2), 1304-1313. https://doi.org/10.1109/78.757218 https://doi.org/10.1016/S1665-6423(14)71612-8 https://doi.org/10.1007/978-90-481-3660-5_10 https://doi.org/10.1016/j.proenv.2011.09.180 https://doi.org/10.5281/zenodo.556248 https://www.ripublication.com/ijaer18/ijaerv13n8_106.pdf https://doi.org/10.1109/ICSIPA.2013.6708032 https://doi.org/10.1049/iet-ipr.2013.0862 https://arxiv.org/ftp/arxiv/papers/1307/1307.0642.pdf https://pdfs.semanticscholar.org/1243/57f41f1b3a87d7380763d2a78041786a35be.pdf https://pdfs.semanticscholar.org/4d0f/50b782f10b01e5e963088d07d25037096263.pdf https://pdfs.semanticscholar.org/4d0f/50b782f10b01e5e963088d07d25037096263.pdf https://doi.org/10.1109/78.757218 THE EFFECT OF TWO MAJOR FACTORS ON THE PERFORMANCE OF STEGANOGRAPHY 51 Copyright ©2019 ASSA. Adv. in Systems Science and Appl. (2019) 18. Tamimi, A. A., Abdalla, A. M., Al-Allaf, O. (2013). Hiding an Image inside another Image using Variable-Rate Steganography, International Journal of Advanced Computer Science and Applications (IJACSA), 4(10), 18-21. https://pdfs.semanticscholar.org/878b/17246ca62e2a62c77fa47c5d6baa0fec3d6f.pdf https://pdfs.semanticscholar.org/878b/17246ca62e2a62c77fa47c5d6baa0fec3d6f.pdf