World Applied Programming, Vol (2), No (3), March 2012. 206-210 ISSN: 2222-2510 2011 WAP journal. www.waprogramming.

com

A Survey on Usage of Genetic Algorithms in Recent Steganography Researches

Masoud Nosrati *
Eslamabad-E-Gharb branch, Islamic Azad University, Eslamabad-E-Gharb, Iran minibigs_m@yahoo.co.uk

Ronak Karimi
Eslamabad-E-Gharb branch, Islamic Azad University, Eslamabad-E-Gharb, Iran rk_respina_67@yahoo.com

Abstract: This paper will take a brief look at the usage of genetic algorithms in steganography. Steganography media which is subjected is image files. So, some of recent studies that introduce the steganography methods are listed and described in brief. As it is seen, robustness and capacity are two important factors that are considered in almost all steganography methods. Also, different methods use one or both of stego-phases. Some of them have some regards about embedding, and some of them modify the stego-image. Keywords: Steganography, data hiding, LSB, DWT. I. INTRODUCTION

Steganography is the art and science to hide data in a cover media such as text, audio, image, video, etc. [1] In other words, steganography is the process of hiding a secret message within a larger one in such a way that someone cannot know the presence or contents of the hidden message [2]. Steganography will hide the message so there is no knowledge of the existence of the message in the place [3]. The term Steganography is forked from the Greek words stegos meaning cover and grafia meaning writing defining it as covered writing [4]. In this paper, steganography in images is subjected. Generally, there are two types of data hiding techniques using images: spatial and frequency domain [5][6]. The first group is based on embedding message in the least significant bit (LSB) of image pixel. The basic LSB method has a simple implementation and high capacity [7]. However it has low robustness versus some attacks such as low-pass filtering and compression [8]. The second group embeds the messages in the frequency coefficients of images. These hiding methods overcome the problem related to robustness and imperceptibility found in the spatial domain. Most recent researches apply discrete wavelet transform (DWT) due to its wide application in the new image compression standard, JPEG2000 [9]. Based on LSB and DWT, many researches is been done. Generally, they tried to increase the robustness of secret data. Robustness measures the ability of embedded data or watermark to withstand against intentional and unintentional attacks. Unintentional attacks generally include common data manipulations such as lossy compression, digital-toanalog conversion, re-sampling, re-quantization, etc. whereas intentional attacks cover a broad range of media degradations which include addition white and colored noise, rescaling, rotation (for image and video steganography schemes), resizing, cropping, random chopping, and filtering attacks [10]. GA is employed to increase the robustness of secret messages against steganalysis [11]. Steganalysis is the process of detecting hidden data which are crested using steganography. Steganalysis detects stego-images by analyzing various image features between stego-images and cover-images [12]. In recent researches, numerous steganography techniques based on genetic algorithms is released. In this study, we are going to have a survey on some of GA-based image steganography researches. II. IMAGE STEGANOGRAPHY METHODS

In this section, some of recent studies and techniques of image steganography will be covered. Methods are briefly described in each part. High capacity image steganography based on genetic algorithm and wavelet transform Elham Ghasemi et al. employ a GA based mapping function to embed data in discrete wavelet transform coefficients in 4*4 blocks on the cover image. The optimal pixel adjustment process (OPAP) is applied after embedding the message. The main idea of applying OPAP is to minimize the error between the cover and the stego-image. Then,

206

Masoud Nosrati et al., World Applied Programming, Vol (2), No (3), March 2012.

utilization of the frequency domain to improve the robustness of steganography is done. Implementation of GA and OPAP is the next step. It is done in order to obtain an optimal mapping function to reduce the difference error between the cover and the stego-image, therefore improving the hiding capacity with low distortions. While the computational complexity is high, simulation results indicate that the novel scheme outperforms adaptive steganography technique based on wavelet transform in terms of peak signal to noise ratio and capacity, 39.94 dB and 50% respectively [9]. Proposed algorithm by Elham Ghasemi et al. for embedding process is listed below [9]: Step 1. Divide the cover image into 4*4 blocks. Step 2. Find the frequency domain representation of blocks by 2D Haar DWT and get four subbands LL1, HL1, LH1, and HH1. Step 3. Generate 16 genes containing the pixels numbers of each 4*4 blocks as the mapping function. Step 4. Embed the message bits in k-LSBs DWT coefficients each pixel according to mapping function. For selecting value of k, images are evaluated from k D 3 to 6. k equal to 1 or 2, provide low hiding capacity with high visual quality of the stego image and k equal to 7 or 8, provide low visual quality versus high hiding capacity. Step 5. Fitness evaluation is performed to select the best mapping function. Step 6. Apply OPAP on the image. Step 7. Calculate inverse 2D-HDWT on each 4*4 block. The extraction algorithm consists of four steps is presented as follow [9]: Step 1. Divide the cover image into 4*4 blocks. Step 2. Extract the transform domain coefficient by 2D HDWT of each 4*4 block. Step 3. Employ the obtained function in the embedding phase and find the pixel sequences for extracting. Step 4. Extract k-LSBs in each pixel.

Data embedding technique for gray scale image using genetic algorithm (DEGGA) J. K. Mandal et al. present another GA-based algorithm termed DEGGA. Focus in this method is on large amount of hidden data and the results are compared with another method by Ran- Zan et al. [13]. In Mandal method, large volume of message/ image is embedded in spatial domain using 3 x 3 masks from the source image. Four bits of the secret message / image is embedded per byte of the source image onto the rightmost 4 bit of each pixel. Mutation is applied on the embedded image. Also, a method of bit handling is applied to keep the fidelity high. In the process of embedding dimension of the secret message / image followed by the content of it. Reverse process is followed during decoding. Genetic algorithm is used to enhance a security level. Various statistical parameters computed that are compared with the Ran- Zan et al. method shows that proposed DEGGA obtained better results in terms of PSNR. Proposed method use gray scale image for secure message transmission. An authenticating image of size m x n is chosen as secret message. The size of the host image is p x q. Input: Host image of size pxq, authenticating image of size pxq. In follow, embedding algorithm is listed. Output of algorithm is embedded image of size pxq. The purpose is inserting the authenticating image (secure message) bitwise into the source image [14]. Embedding algorithm [14]: Step 1. Obtain the size of the authenticating image m x n. Step 2. For each authenticating message/image, Read source image block of size 3x3 in row major order. Extract authenticating message/image bit one by one. Replace the authenticating message/image bit in the rightmost 4 bits within the block, four bits in each byte. Step 3. Read one character/ pixel of the authenticating message/ image at a time. Step 4. Repeat step 2 and 3 for the whole authenticating message/ image size, content. Step 5. Perform mutation operation for the whole embedded image. For mutation rightmost 3 bits from each bytes is taken. A consecutive bitwise XOR is performed on it for the 3 steps. It will form a triangular form and first bit from each step is taken. Step 6. A bit handling method is performed on the embedded image. If the difference between the host and embedded image is 16, then 16 will be added to the embedded image to keep intact the visibility of the embedded image. Step 7. Stop. Extraction algorithm [14]: Embedded image of size pxq is the input of extraction algorithm. In other hand, output will be the host image of size pxq and authenticating image of size pxq.

207

Masoud Nosrati et al., World Applied Programming, Vol (2), No (3), March 2012.

Step 1. Step 1: Perform reverse mutation procedure by consecutive bitwise XOR operation on the rightmost 3 bits of each byte in three steps. The first bit of each step is taken as the output. Step 2. Step 2: Read the image block of size 3x3 in row major order. Step 3. Step 3: For each block, extract the authenticating image from the rightmost 4 bits of each byte replace authenticating message/ image bit position in the block by '1'. For each 8 (eight) bits extraction construct one character/ one image pixel. Step 4. Step 4: Repeat step 2 and 3 to complete decoding as per size of the authenticating image. Step 5. Step 5: Stop

Secure steganography method based on genetic algorithm To ensure the security against the RS analysis, Shen Wang et al. present a new steganography method based on genetic algorithm. There is no approach for embedding data in host image in proposed method. But, securing the stego-image is done after data embedding. After embedding the secret message in LSB (least significant bit) of the cover image, the pixel values of the stego-image are modified by the genetic algorithm to keep their statistic characters. Thus, the existence of the secret message is hard to be detected by the RS analysis. Meanwhile, better visual quality can be achieved by the proposed algorithm. The experimental results demonstrate the proposed algorithm's effectiveness in resistance to steganalysis with better visual quality [15]. In RS steganalysis, 3 kinds of block flipping are defined. They are positive flipping F1, negative flipping F-1 and 0 flipping F0. F1 is the transformation relationship between 2i and 2i+1, (i.e. 0-1, 2-3, , 254-255), which is same as LSB. F-1 is the transformation relationship between 2i-1 and 2i, i.e. -1-0, 1-2, , 255-256. The relationship between the two flipping is written as: F-1 = F1(x + 1) - 1 Similarly, define F0 as the identity permutation F0(x) = x F0, F1 and F-1are called flipping functions. The flipped group is resulted from applying flipping functions on pixels of image block. It is denoted as: F(G) = (FM(1)(x1); FM(2)(x2); : : : ; FM(n)(xn)) M = M(1);M(2); : : : ;M(n) is called a flipping mask, where M(i) are 1, 0, or -1. G is regular if f(F(G)) > f(G), G is singular if f(F(G)) < f(G). The RS analysis includes following steps. First, the image is divided into non-overlapping blocks and each one is rearranged into a vector G = (x1, x2, , xn) in the Zigzag scan order. The correlations of pixels can be determined by discrimination function:

Where x is the pixel value and n is the number of pixels. The value of f represents the spatial correlation between the adjacent pixels. A small f means the strong correlation. After all the f(G) are obtained, apply non-negative flipping (i.e. M(1);M(2); : : : ;M(n) = 0 or 1) and non-positive flipping (i.e. M(1);M(2); : : : ;M(n) = 0 or -1) on each block. Then calculate f(F(G)) in each block. The relative number of regular blocks after positive flipping is denoted as Rm, and that of singular blocks is denoted as Sm. In the same way, R-m and S-m are defined as the relative number of regular and singular blocks after the negative flipping. It is pointed out that in nature images, the numbers of aforementioned blocks hold the following relationships: Rm R-m, Sm S-m and Rm > Sm,R-m > S-m The difference between Rm and R-m increases with the length of the embedded message. The same trend exists in the difference between Sm and S-m. In Shen Wang et al. study genetic algorithm is used to search for a best adjustment in order to make stego-image undetectable by RS [15].

208

Masoud Nosrati et al., World Applied Programming, Vol (2), No (3), March 2012.

Improving the performance of LSB substitution against RS by minimizing detection probability One of the most notable steganalysis algorithms is the RS attack which detects the stego-message by the statistic analysis of pixel values [16]. Another steganographic algorithm is presented by Vijay Kumar Sharma et al. based on genetic algorithm. In this method, after embedding the secret message in LSB (least significant bit) of the cover image, the pixel values of the stego image are modified by the genetic algorithm to keep their statistic characters. For modification of stego image, the method of Shen Wang et al. [14] is used. Thus, the existence of the secret message is hard to be detected by the RS analysis. Meanwhile, better visual quality can be achieved by the proposed algorithm. The experimental results demonstrate the proposed algorithm's effectiveness in resistance to steganalysis with better visual quality [17]. For this purpose we employ the Genetic Approach: Proposed GA is as follow: Step 1. Initialization: Starting from the first pixel, select two pixels adjacent to each other in a row as shown in the figure 1. These are known as the initial chromosomes in the genetic terminology. Step 2. Reproduction and Mutation: Flip the second lowest bits in the chromosomes randomly to generate the four possible chromosomes of the next generation as shown in figure 2. Step 3. Selection: Select the best possible chromosome which maximizes the value of the flipping function F=.Prob[f-(Ci)<f(Ci)]+.PSNR Where and are positive constants. Also the Probability Prob[f-(Ci)<f(Ci)] must be greater than some threshold. Step 4. Calculate the label of the adjusted image block using genetic algorithm. The block is successfully adjusted. Step 5. Crossover. Shift the chromosomes one pixel, goto step 2. If crossover has been applied two times, stop the cycle. Step 6. After a block is adjusted, calculate R ,R, S and S of the image. If the difference between R and R is more than 5%, or the difference between S and S is more than 5%, adjust the next block.

Figure 1. Selection of two pixels adjacent to each other in a row for initial state.

Figure 2. Four possible chromosomes of the next generation. Some optimizations in the proposed method caused reducing the computational complexity of the algorithm. Also, implementation of this method seems to be easy [17].

Improved adaptive LSB steganography based on chaos and genetic algorithm Lifang Yu et al. present a steganography method in JPEG images with high performance. Proposed method is consisting of 2 parts. First, improved adaptive LSB steganography is presented, which can achieve high capacity while preserving the first-order statistics. Second, in order to minimize visual degradation of the stego image, shuffling bits-

209

Masoud Nosrati et al., World Applied Programming, Vol (2), No (3), March 2012.

order of the message based on chaos is done, whose parameters are selected by the genetic algorithm. The chaos phenomenon is a deterministic and analogously stochastic process appearing in a nonlinear dynamical system [18][19]. Because of its extreme sensitivity to initial conditions and the outspreading of orbits over the entire space, it has been used in information hiding to increase security [20][21]. Experimental results show that our method outperforms classical steganographic methods in image quality, while preserving characteristics of histogram and providing high capacity [22]. III. CONCLUSION

Steganography in image files uses 2 general techniques: LSB (Least Significant Bits) and DWT (Discrete Wave Transform). Proposed methods of steganography use one of these techniques. Robustness is the main consideration in steganography. Furthermore, capacity is considered as an important factor, too. Stego methods concentrate on these factors and commonly there make a sensible proportion between them. Steganography methods commonly guarantee the data security in 2 phases: First, some of methods concentrate on manner of embedding data in host image. For example, "Data embedding technique for gray scale image using genetic algorithm (DEGGA)" method focuses on place of data on stego-image, while in "Secure steganography method based on genetic algorithm" there is no focus on manner of embedding. Second, some methods modify the stego-image after embedding secret message. For example, "Secure steganography method based on genetic algorithm" method has no regards about data embedding, but it guarantees the secret message by modifying the stego-image in order to bypass the RS test. Also, some of methods use both of phases like "High capacity image steganography based on genetic algorithm and wavelet transform". REFERENCES
[1] [2] [3] [4] [5] [6] [7] [8] [9] Inoue H, Miyazaki A, Katsura T (2002) An image watermarking method based on the wavelet transform. In: International Conference on Image Processing, IEEE ICIP, Kobe, vol 1, pp 296300. Masoud Nosrati, Ronak Karimi, Mehdi Hariri. Embedding Stego-Text in Cover Images Using linked List Concepts and LSB Technique. World Applied Programming, Vol (1), No (4), October 2011. 264-268. Masoud Nosrati, Ronak Karimi, Mehdi Hariri. An introduction to steganography methods. World Applied Programming, Vol (1), No (3), August 2011. 191-195. S. Katzenbeisser, F.A.P. Petitcolas, Information Hiding Techniques for Steganography and Digital Watermarking, Artech House, Norwood, MA, 2000. Provos N, Honeyman P (2003) Hide and seek: an introduction to steganography. In: IEEE Journal Computer Society, ISSN: 1540-7993, 11 June 2003, pp 3244 Provos N (2001) Defending against statistical steganalysis. In: SSYM01 Proceedings of the 10th conference on USENIX Security Symposium, USENIX Association Berkeley, CA, USA, vol 10, pp 323335. Wu N, Hwang M (2007) Data hiding: current status and key issues. Int J Network Security 4(1):19. Chen W (2003) A comparative study of information hiding schemes using amplitude, frequency and phase embedding, PhD thesis, National Cheng Kung University, Taiwan. Elham Ghasemi, Jamshid Shanbehzadeh, and Nima Fassihi. High Capacity Image Steganography Based on Genetic Algorithm and Wavelet Transform, S.I. Ao et al. (eds.), Intelligent Control and Innovative Computing, Lecture Notes in Electrical Engineering 110, DOI 10.1007/9781-4614-1695-1 30. Cvejic N. and Seppnen T. Increasing the capacity of LSB based audio steganography, Proc. 5th IEEE International Workshop on Multimedia Signal Processing, St. Thomas, VI, December 2002, pp. 336-338. ElShafie DR, Kharma N, Ward R (2008) Parameter optimization of an embedded watermark using a genetic algorithm. In: International symposium on communications, control and signal processing, ISCCSP, St Julians1214 March, pp 12631267 Frank Y. Shih, Yi-Ta Wu. Digital Steganography Based on Genetic Algorithm. Handbook of Research on Secure Multimedia Distribution, DOI: 10.4018/978-1-60566-262-6.ch023. Ran-Zan Wang, Chi- Fang Lib, and Ja- Chen Lin, "Image hiding by optimal LSB substitution and Genetic algorithm", 2001 Pattern Recognition Society. Published by Elsevier Science Ltd. J. K. Mandal, A. Khamrui. A Data Embedding Technique for Gray scale Image Using Genetic Algorithm (DEGGA). International Confrence on Electronic Systems (ICES-2011). Shen Wang, Bian Yang and Xiamu Niu A Secure Steganography Method based on Genetic Algorithm Journal of Information Hiding and Multimedia Signal Processing c 2010 ISSN 2073-4212. Volume 1, Number 1, January 2010 J. Fridrich, R. Du, and L. Meng, Steganalysis of LSB Encoding in Color Images, Proc. IEEE International Conf. Multimedia and Expo, CDROM, IEEE Press, Piscataway, N.J., 2000. Vijay Kumar Sharma, Vishal Shrivastava. Improving the Performance of Least Significant Bit Substitution Steganography against Rs Steganalysis by Minimizing Detection Probability. International Journal of Information and Communication Technology Research, Volume 1 No. 4, August 2011. Z. Liu and L. Xi, Image information hiding encryption using chaotic sequence, in Proceedings of the 11th International Conference on Knowledge-Based Intelligent Information and Engineering Systems and the XVII Italian Workshop on Neural Networks, pp. 202208, 2007. Y. Zhang, F. Zuo, Z. Zhai, and C. Xiaobin, A new image encryption algorithm based on multiple chaos system, in Proceedings of the International Symposium on Electronic Commerce and Security (ISECS 08), pp. 347350, August 2008. R. Munir, B. Riyanto, S. Sutikno, and W. P. Agung, Secure spread spectrum watermarking algorithm based on chaotic map for still images, in Proceedings of the International Conference on Electrical Engineering and Informatics, 2007. Z. Dawei, C. Guanrong, and L.Wenbo, A chaos-based robust wavelet-domain watermarking algorithm, Chaos, Solitons and Fractals, vol. 22, no. 1, pp. 4754, 2004. Lifang Yu, Yao Zhao, Rongrong Ni, Ting Li. Hindawi Publishing Corporation, EURASIP Journal on Advances in Signal Processing, Volume 2010, Article ID 876946. doi:10.1155/2010/876946.

[10] [11] [12] [13] [14] [15] [16] [17]

[18] [19] [20] [21] [22]

210