A Real Time Image Steganalysis by Chi-Square Test (CTSI) Method

A Real Time Image Steganalysis by Chi-Square Test (CTSI) Method Sabyasachi Samanta 1, Saurabh Dutta 2, Goutam Sanyal 3 1 Haldia Institute of Technology, Haldia, WB, INDIA E-mail id: sabyasachi.smnt@gmail.com 2 Dr. B. C. Roy Engineering College, Durgapur, WB, INDIA E-mail id: saurabh.dutta@bcrec.org 3 National Institute of Technology, Durgapur, WB, INDIA E-mail id: nitgsanyal@gmail.com Abstract- Steganography is the art of hiding data in data in an untraceable way. Main concern of steganography is hiding the existence of hidden message. Steganalysis is the art and science of detecting hidden messages from stego-systems. It also attempts to find hidden message such as the type of embedding algorithm, the length of the message, the content of the message or the secret key used from the carrier (like image, text, video etc.). In this paper we have discussed about the steganalysis by using Chi-square Test based Steganalysis for Image (CTSI) Method. Keywords: Steganography, Steganalysis, Cover image, Stego image, Chi-square Test based Steganalysis for Image (CTSI) Method I. INTRODUCTION Steganography is the art of hiding data in data in an undetectable way. Steganography related with cryptography but it is different [3] [4]. Main concern in cryptography is hiding the content of the message but in steganography main concern is hiding the existence of hidden message. Steganalysis is the art and science of detecting hidden messages from images, text etc. made from stego-systems [21]. Steganalysis is a fast growing science and relatively new extent. The purpose of steganalysis is to distinguish if an image contains a secret message or not [5] [6]. Steganalysis is based on the combination of stego-media, embedded message and steganography tools known by the steganalyst [17][20]. In this section, basic about types of attacks used by the steganalyst [7], different approaches of steganalysis [7] and Chisquared (x 2 ) Test [14] [15] have been discussed. A) Types of attacks used by the steganalyst [18]: Stego-Only Attack: In stego-only attack only the stego-object is available for analysis. Known Cover Attack: The original cover-object is compared with the stego object in this type of attack. Known Message Attack: In known message attack the attacker may know the hidden message. Chosen Stego Attack: The steganography algorithms and stego-object are known in this type of attack. Chosen Message Attack: The steganalyst generates a stego-object from some steganography algorithm using a chosen message. Known Stego Attack: The steganography tool is known and both the original and stego-object are available. B) Different Approaches of Steganalysis Three types of approaches are there [16] [19]: I. Visual Attacks: Visual attack mostly involves examining the content file with the naked eye to identify any noticeable inconsistencies. The most important rule of steganography is that any modification made to content should not result in quality degradation. A good steganographic implementation creates a stego-object which is not suspicious than the cover work. II. Structural Attacks: RES Publication 2012 Page 25

Structural attack is on size of the file. The layout of carrier file habitually changes as the data to be hidden within it. That characteristic of structural changes can detect the existence of hidden message with in the carrier. Structural attack is arguably more important to steganalysts than visual attack. III. Statistical Attacks: The idea behind the statistical attack [1] is to compare the theoretically expected frequency distribution in stego carrier with some sample distribution observed in the possibly changed carrier medium. Pfitzman and Westfield [8] introduced the powerful statistical attack which can be applied to any steganographic technique by which fixed set of Pairs of Values (PoVs) are flipped into each other at the time of embedding the message bits. C) Chi-squared (x2) Test One of the simplest and most popular statistical attacks is the Chi-squared Test (also referred to as the x 2 Test) which was documented in steganalytical term by Andreas Westfeld and Andreas Pfitzmann in the year of 1999 []. The test makes it possible to compare the statistical properties of a suspect image with the theoretically expected statistical properties of its carrier counterpart such that it is possible to determine the likelihood that a suspect image is a stegogramme. When we overwrite on any significant bit position of an image with data from a message bit stream, we transform the values into each other. Let us consider an image where the first pixel value = 194. If we embed a 0, the value will remain the same. If we embed a 1 to LSB position, the pixel value will change to 195. If anybody change bit to upper LSB position(s) the pixel value changes. The formation of odd and even value to be pairs, known in steganalysis as Pairs of Values (PoVs), where the PoVs for an 8-bit gray scale image would involve of pixel values {0, 1; 2, 3; :::; 252, 253; 254, 255}. In case of color image the combination of PoVs may be consider for each of the RGB components. So the expected distribution of the sum of neighboring values for the image to be:... (i) Now that we have effectively obtained the expected distribution for the cover work from equation (i), we can compare the result against that of the suspect image by using the x 2 formulae in equation (ii) with (v 1) degrees of freedom: Substituting equation (i) into equation (ii) directly produces:... (ii).. (iii) Section 2 represents the related work. Section 3 represents an implementation of the technique. Section 4 gives you an idea about the experimental results. Section 5 is an analytical discussion on the technique. Section 6 draws a conclusion. II. RELATED WORK In this section we have studied a number of earlier methodologies. Westfeld and Pfitzmann [8] propose an approach which is specific to LSB embedding and is based on powerful first order statistical analysis. It detects Pairs of Values (POVs) that consist of pixel values. After embedding the message, the total number of occurrence of two members of certain POV remains the same. This concept of pair wise dependencies is leads to design a statistical Chi-square test to detect the hidden messages [9]. Zhang and Ping [10] propose a technique uses different image histogram as the statistical analysis tool. Measure of the correlation between the LSB plane and the rest of the planes is done by the translation coefficients between different RES Publication 2012 Page 26

image histograms. This algorithm can identify the existence of secret messages embedded using sequential or random LSB replacement in images and also can estimate the amount of secret messages. Kenneth Sullivan et al. [11] motivated on the detection of hidden data. They have used Markov chain image model, statistical analysis of spread spectrum hiding under this model, and estimations of the detectability of various adaptations of SS hiding specially. They focused on detecting data hidden in grayscale images with spread spectrum hiding. Chandramouli and Subbalakshmi [12] proposed two steganalysis schemes specifically for spread spectrum steganography. First scheme is a simple to estimate and subtract type algorithm which does not exploit higher order statistics. Assessment of cover image from stego image is done by standard regression techniques. The estimated value is subtracted from the stego image to get the estimate of the covert message [25]. Roshidi Din, T. Zalizam T. Muda and coauthors [13] proposes a scheme namely Evolution Detection Steganalysis System (EDSS) based on the evolution algorithm approach under Java Genetic Algorithms Package for text steganalysis. III. THE SCHEME Here the stego-image is taken, where the data is embedded to any of last four bit positions. Here we have used the known cover attack method to detect the presence of message in carrier. Initially the histogram separately with RGB values for each original and stego image have been drawn. Similarly, the histogram of RGB for each cover and stego-image with PoV has been drawn. Also the graph for pixel difference of histogram attack has been drawn. IV. EXPERIMENTAL RESULT A LENA image with 256x256 dimensions has been taken. We have embedded 100 characters to that image and get stego image. (a) (b) (c) Figure 4.1: (a) (b) and (c) Graphical Representation of RGB Values for LENA Cover Image Respectively (a) (b) Figure 4.2: (a) (b) and (c) Graphical Representation of RGB Values for LENA Stego-Image Respectively RES Publication 2012 Page 27

(a) (b) Figure 4.3: (a) (b) and (c) Graphical Representation of RGB Difference Values for LENA Cover Image Respectively (a) (b) Figure 4.4: (a) (b) and (c) Graphical Representation of RGB Pair Difference Values for LENA Stego-Image Respectively Figure 4.5: Chi square analysis for LENA image Figure 4.6: Result for Pixel Difference of Histogram Attack V. ANALYSIS Here in this paper we have discussed image based steganalysis scheme for breaking steganography. Here we have uses the known cover attack method to detect the presence of message in carrier. After embedding only 100 characters, we get some visual dissimilarity of RGB graph with CTSI statistical analysis. But at the time of common appearance of RGB histogram, we do not get that type of dissimilarity. With embedding of more number of characters to cover image, the newly developed methodology gives superior results. RES Publication 2012 Page 28

VI. CONCLUSION Here we have developed statistical image steganalysis tool based on Chi-squared (x 2 ) Test methodology. Optimistically, this study may be enabling for future research and to develop better steganalysis tools that can contribute to better performance. REFERENCES [1] Udit Budhia, Deepa Kundur, Takis Zourntos, Digital Video Steganalysis Exploiting Statistical Visibility In The Temporal Domain, IEEE Transactions on Information Forensics and Security, Vol. 1, No. 4, December 2006, pp. 502-516 [2] Fengyong Li, Xinpeng Zhang, Bin Chen, Guorui Feng, JPEG Steganalysis with High-Dimensional Features and Bayesian Ensemble Classifier, IEEE Signal Processing Letters, Vol. 20, No. 3, March 2013, pp. 233-236 [3] Chunfang Yang, Fenlin Liu, Xiangyang Luo, Ying Zeng, Pixel Group Trace Model-Based Quantitative Steganalysis for Multiple Least-Significant Bits Steganography, IEEE Transactions on Information Forensics and Security, Vol. 8, No. 1, January 2013, pp.216-218 [4] Jessica Fridrich, Jan Kodovský, Rich Models for Steganalysis of Digital Images, IEEE Transactions on Information Forensics and Security, Vol. 7, No. 3, June 2012, pp. 868-882 [5] Gokhan Gul, Fatih Kurugollu, JPEG Image Steganalysis Using Multivariate PDF Estimates with MRF Cliques, IEEE Transactions on Information Forensics and Security, Vol. 8, No. 3, March 2013, pp.578-587 [6] Bin Li, Junhui He, Jiwu Huang, Yun Qing Shi A Survey on Image Steganography and Steganalysis, Journal of Information Hiding and Multimedia Signal Processing, Volume 2, Number 2, April 2011, pp. 142-172 [7] Pierre Richer, Steganalysis: Detecting hidden information with computer forensic analysis, SANS/GIAC Practical Assignment for GSEC Certification, Version 1.4b [8] A. Westfeld, A.Pfitzmann, Attacks on steganographic systems, Proc. of Information Hiding, Third Int. Workshop, Dresden, Germany, September 28 October 1, 1999, pp. 61 75. [9] N.F. Johnson, S. Jajodia, Steganalys of images created using current steganography software, in: Lecture Notes in Computer Science, vol. 1525, Springer-Verlag, Berlin, 1998, pp. 273 289. [10] T. Zhang, X. Ping, Reliable detection of LSB steganography based on difference image histogram, in: Proc. ICASSP, vol. I, 2003, pp. 545 548. [11] Kenneth Sullivan, Upamanyu Madhow, Shivkumar Chandrasekaran, and B.S. Manjunath, Steganalysis of Spread Spectrum Data Hiding Exploiting Cover Memory http://vision.ece.ucsb.edu [12] R. Chandramouli, K.P. Subbalakshmi, Active steganalysis of spread spectrum image steganography, IEEE Int. Symp. on Circuits and Systems, Bangkok, Thailand, vol. 3, May 2003, pp. 830 833. [13] Roshidi Din, T. Zalizam T. Muda, Puriwat Lertkrai, Mohd Nizam Omar, Angela Amphawan, Fakhrul Anuar Aziz, Text Steganalysis Using Evolution Algorithm Approach, Advances In Computer Science, ISBN: 978-1-61804-126-5, Pp. 444-449 [14] Vicki Sharp, Chi-Square Test", Statistics for the Social Sciences [15] Jennifer L. Waller, How to Perform and Interpret Chi-Square and T-Tests, Paper 155 SAS Global Forum 2012, Hands-on Workshops [16] Mrs. Kavitha, Kavita Kadam, Ashwini Koshti, Priya Dunghav, Steganography Using Least Significant Bit Algorithm, International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622, Vol. 2, Issue 3, May-Jun 2012, pp. 338-341 [17] Natarajan V1, R Anitha Blind Image Steganalysis Based on Contourlet Transform, International Journal on Cryptography and Information Security (IJCIS),Vol.2, No.3, September 2012, pp.77-87 [18] Souvik Bhattacharyya, Gautam Sanyal, Moments and Similarity Measure Feature Based Image Steganalysis Technique (MSM), International Journal of Information & Network Security (IJINS), Vol.2, No.2, April 2013, pp. 138-153 [19] Natarajan Meghanathan, Lopamudra Nayak, Steganalysis Algorithms for Detecting the Hidden Information in Image, Audio and Video Cover Media International Journal of Network Security & Its Application (IJNSA), Vol.2, No.1, January 2010, pp. 43-55 [20] Wien Hong and Tung-Shou Chen, A Novel Data Embedding Method Using Adaptive Pixel Pair Matching, IEEE Transactions on Information Forensics and Security, Vol. 7, No. 1, February 2012, pp. 176-184 [21] Sabyasachi Samanta, Saurabh Dutta, Gautam Sanyal, Enhancement of Security of Data through a Combined Approach of Encryption and Watermarking International Conference on Information and Mathematical Sciences, October, 2013, pp. 342-345 RES Publication 2012 Page 29

[22] R. Chandramouli, A Mathematical Approach to Steganalysis, Multimedia Systems, Networking and Communications (MSyNC) Lab Department of Electrical and Computer Engineering, Stevens Institute of Technology [23] Jonathan Dautrich, Multi-Class Steganalysis, Machine Learning, UC Riverside, Professor Shelton June 12, 2009 [24] Jessica Fridrich, Jan Kodovský, Rich Models for Steganalysis of Digital Images, IEEE Transactions on Information Forensics and Security, Vol. 7, No. 3, June 2012, pp. 868-882 [25] R. Chandramouli and K.P. Subbalakshmi, Current Trends in Steganalysis: A Critical Survey, Department of ECE, Stevens Institute of Technology RES Publication 2012 Page 30