BLIND EXTRACTION OF HIDDEN DATA FROM DIGITAL IMAGE USING M-IGLS ALGORITHM D.PADMAVATHI 1 (PG SCHOLAR) DR.K.BABULU 2 PH.D 1 Department of Electronics and Communications, JNTU Kakinada, AP, INDIA 2 Professor, Department of Electronics and Communications, JNTU Kakinada, AP, INDIA ABSTRACT Data hiding and extraction schemes are increasing in today s communication world due to rapid increment of data tracking and tampering attacks.so we need an efficient and robust data hiding schemes to protect from these attacks In this project the blindly extraction technique is considered. Blindly extraction means the original host and the embedding carriers are not need to be known. Here, the, hidden data embedded to the host signal, via multicarrier SS embedding. ). Implementation of hiding data in image data using Direct Sequence Spread Spectrum method has been presented in this work. The hidden data is extracted from the digital media like audio, video or image. The extraction algorithm used to extract the hidden data from digital media is Multicarrier Iterative Generalized Least Squares (M-IGLS).It is a low complexity algorithm and it attains the probability of error recovery equals to known host and embedding carriers. Its peak signal to noise ratio value obtained is high. Experimental studies on images show that the developed algorithm can achieve recovery probability of error close to what may be attained with known embedding carriers and host autocorrelation matrix. Keywords: Data hiding, Tracking, Tampering, Blindly extraction, Spread spectrum embedding. 1. INTRODUCTION In the field of Data Communication, security-issues have the major problem. Data tracking and tampering are rapidly increasing in everywhere like online tracking, mobile tracking etc. So we need a Secured communication scheme for transmitting the data. For that, we are having many data hiding schemes and extraction schemes. Hiding the information is a vital issue in the 21 st century in the field of Data Communication security.its an important issue because the virtual and digital information transmission faces critical setbacks due to hacking and hackers threats. The transmission of information via the Internet may expose it to detect and theft. So data embedding technologies are developed to provide personal privacy, commercial and national security interests. Digital data embedding in digital media is rapidly growing commercial as well as national security interest. Data hiding schemes are initially used in military communication systems like encrypted message, for finding the sender and receiver or its very existence. Initially the data hiding schemes are used for the copy write purpose. In [1] Fragile watermarks are used for the authentication purpose, i.e. to find whether the data has been altered or not. Likewise the data extraction schemes also provide a good recovery of hidden data.this is the goal of the secured communication. The main Applications of data hiding are annotation, copyright-marking, and watermarking, singlestream media merging (text, audio, image) and Steganography [1].The proposed paper having the blindly extraction technique is considered. Blindly extraction means the original host and the embedding carriers are not require to be known. Here, data embedded to the host digital signal, via multicarrier spread spectrum embedding. Hence it has developed rapidly in this area due to the advantages of good robustness and immunity to noise attack. Spread spectrum techniques really of digital communications systems. Two commonly spread spectrum techniques are used direct sequence spread spectrum (DSSS) and frequency hopped spread spectrum (FHSS). Implementation of hiding data in image data using Direct Sequence Spread Spectrum method has been presented in this work. Hidden data is extracted from the digital media like audio, video or image. The extraction algorithm used to extract the hidden data from digital media is Multicarrier Iterative Generalized Least Squares (M-IGLS). This blind hidden data extraction problem has also been referred to as Watermarked content Only Attack (WOA) in the watermarking security context [2] [5].
\ In blind active spread spectrum detection is the unknown host acts as source of interference/disturbance to the data to be extracted with, in a way; the trouble parallels blind signal separation (BSS) applications as they arise in the fields of biomedical signal processing, array processing and code division multiple access (CDMA) communication systems. In proposed paper, we implement a new multisignature iterative generalized least squares (M-IGLS) SS algorithm for hidden data extraction. For improved recovery performance and in particular for small hidden messages that pose the greatest challenge, we propose an algorithmic upgrade referred to as cross-correlation enhanced M-IGLS (CC-M-IGLS). CCM-IGLS relies on statistical analysis of independent M-IGLS executions on the host and experimental studies demonstrate hidden data recovery with probability of error close to what may be attained with known embedding signatures and known original host autocorrelation matrix are implemented. 2. RELATED WORK There are many data hiding and data extraction schemes are comes into existence. The important data hiding technique is watermarking. It is differ from steganography and cryptography in the way of data hiding. In Steganography a secret message is hidden within another unrelated message and then Communicated to the other party. As opposed to this in Watermarking again one message is hidden in another, but two messages are related to each other in some way. Cryptography only provides security by encryption and decryption so the data unreadable by a third party. In [6] there are many extraction procedures to seek the hidden data. But it is having some disadvantages. Iterative Least Square Estimation (ILSE) is prohibitively complex even for moderate values. Pseudo-ILS (ILSP) algorithm is not guaranteed to converge in general and also it provides measurably worse results. So, these two algorithms coupled and so called Decoupled weighted ILSP(DW-ILSP).But here also have an disadvantage like,it may not be valid for large N. So we going for M-IGLS extraction algorithm for extraction of hidden data from digital image. In my project the concept of Watermarked Content only attack in the watermarking security context is taken.i.e. the blindly recovery of data is considered. In this Spread-spectrum-like discrete cosine transform domain (DCT domain) technique of digital images is analyzed. The DCT is applied in blocks of 8 8 pixels as in the JPEG algorithm. The watermark can encode information to track illegal misuses. For flexibility purposes, the original image is not necessary during the ownership verification process, so it must be modeled by noise. Two tests are involved in the ownership verification stage: watermark decoding, in which the message carried by the watermark is extracted, and watermark detection, which decides whether a given image contains a watermark generated with a certain key. We apply generalized Gaussian distributions to statistically model the DCT coefficients of the original image and show how the resulting detector structures lead to considerable improvements in performance with respect to the correlation receiver, which has been widely considered in the literature and makes use of the Gaussian noise assumption. A family of DCT-domain watermarking techniques based on spread spectrum modulation schemes has been analyzed. Security and robustness are provided by means of a secret key whose value determines the output of the pseudorandom sequence generator and the pseudorandom sample permutation performed in the interleaving stage. The original image is not necessary during the watermark detection and decoding processes, for the sake of flexibility in the possible applications of the watermarking process. A statistical characterization of the original image s DCT coefficients has been provided, using the generalized Gaussian model. Then, optimalml detector structures have been derived for both the watermark decoding (extraction) and watermark detection problems. In [7] spread spectrum embedding algorithm for blind watermarking have based on the understanding that the host signal acts as a source of interference to the secret message of interest. Such knowledge can be useful for the blind receiver at the recovery side to minimize the recovery error rate for a given host signal. To increase the security and payload rate the embeder will take multicarrier embedding concept. In [8] the spread spectrum communication is explained. Here a narrow band signal is transmitted over a much larger bandwidth such that the signal energy present in any single frequency is imperceptible. Similarly in SS embedding scheme, the hidden data is spread over many samples of host signal by adding a low energy Gaussian noise sequence. The DCT transformation is taken for embedding purpose as a carrier since it is a fast algorithm and for it s efficient implementation. In [9] the Generalized Gaussian Distribution (GGD) has been used to model the statistical behavior of the DCT coefficients
INTERNATIONAL JOURNAL
INTERNATIONAL JOURNAL
4 SIMULATION RESULTS Quality of a hidden message extraction solution is the difference in bit-error-rate (BER) experienced by the intended recipient and the analyst. The intended recipient using the (i) ideal MMSE filtering with known carriers and known true host autocorrelation matrix In terms of blind extraction (neither nor known), we will examine: (ii) M-IGLS algorithm in Table I with P=20 reinitializations. 2) First consider a gray-scale host image 512 512 Baboon as shown in fig1. We perform 8 8 block DCT embedding and consider all bins except the dc coefficient with K=4 hidden messages and distinct arbitrary carriers. length of embedding carriers is L=63.Each message is having a length of 4K bits. The per- message block mean square distortion due to each embedded message is set to be the same for all messages. The graph between average BER versus per-message block distortion is shown in fig2.
Fig.1. 512*512 baboon image Fig.3.256*256 baboon image.4. average BER versus per message block distortion L=63, K=4 messages of 1K bits each Comparing with Fig. 2, the gap between M-IGLS and ideal MMSE increases as the hidden message size decreases, but the extraction performance of M-IGLS seems to be satisfactory. Fig.2. average BER versus per message block distortion 1) Similar experimentation is done for 512 512 gray scale boat image.fig5 shows the 512 512 boat image. Embedding is 8 8 DCT with K=4 messages each of length 4K bits and embedding carriers of length L=63. L=63, K=4 messages of 4K bits each From fig.2. We can say that M-IGLS analysis is very close in BER performance to the ideal MMSE detector. 1) Now let us consider 256 256 gray scale baboon image as shown in fig.3. With K=4 hidden messages each of length 1K bits with same L=63. The graph between average BER versus per message block distortion for fig.3 is shown in fig.4. And see how it is going to differ from fig.2. Fig.5. 512*512 boat image The per- message block mean square distortion due to each embedded message is set to be the same for all messages. The graph between average BER versus per-message block distortion is shown in fig.6.
L=63, K=4 messages of 1K bits each Fig.6. average BER versus per message block distortion L=63, K=4 messages of 4K bits each Fig.9. 512*512 plane image 1) Now let us consider 256 256 gray scale boat image as shown in fig.7. With K=4 hidden messages each of length 1K bits with same L=63. The graph between average BER versus per message block distortion is shown in fig.8. Fig.7. 256*256 boat image distortion Fig.10. average BER versus per message block L=63, K=4 messages of 4K bits each Fig.8. average BER versus per message block distortion Fig.11. 256*256 plane image
REFERENCES [1]F.A.P.Petitcolas, R.J.Anderson, and M.G.Kuhn. Information hiding. A survey,,proc. IEEE, Special Issue on Identification and Protection of Multimedia Information, vol. 87, no. 7, pp. 1062 1078, Jul. 1999. [2] F. Cayre, C. Fontaine, and T. Furon, Watermarking security: Theory and practice, IEEE Trans. Signal Process., vol. 53, no. 10, pt. 2, pp. 3976 3987, Oct. 2005. Fig.12. average BER versus per message block distortion L=63, K=4 messages of 1K bits each Conclusion from all these experiments is that M- IGLS is most effective technique to blindly extract hidden messages, while extraction becomes more challenging as the length of the hidden message per used embedding carrier decreases or the number of hidden messages (number of used carriers) increases. 4. CONCLUSION In this project multicarrier iterative generalized least-square (M-IGLS) core algorithm has been proposed for blind extraction of unknown data hidden in images via DCT transform by multicarrier spread spectrum embedding. This algorithm is having a low complexity. This extraction technique will provides high peak signal to noise ratio and it will attains the probability of error recovery equals to known host and embedding carriers. This technique is enhanced by using harmony search algorithm where it provides low time consumption and high attack resistance. Blind data extraction algorithmic development was based on the most common SS embedding, the developed algorithm can also be applied to more advanced SS embedding schemes such as improved spread-spectrum (ISS) and correlation-aware improved spread-spectrum (CAISS). ACKNOWLEDGEMENT I am very thankful to our Prof. Dr. K. Babulu for his consistent support and guidance throughout the project period. I acknowledge the support of my peers who supported me in this work. [3] L. Pérez-Freire, P. Comesana, J. R. Troncoso-Pastoriza, and F. Pérez-González, Watermarking security: A survey, LNCS Trans. Data Hiding Multimedia Security, vol. 4300, pp. 41 72, Oct. 2006. [4] M. Barni, F. Bartolini, and T. Furon, A general framework for robust watermarking security, ACM J. Signal Process., Special Section: Security of Data Hiding Technologies, vol. 83, pp. 2069 2084, Oct. 2003. [5] L. Pérez-Freire and F. Pérez-González, Spreadspectrum watermarking security, IEEE Trans. Inf. Forensics Security, vol. 4, no. 1, pp. 2 24, Mar. 2009. [6] T. Li andn.d. Sidiropoulos, Blind digital signal separation using successive interference cancellation iterative least squares, IEEE Trans Signal Process., vol. 48, no. 11, pp. 3146 3152, Nov. 2000. [7]. M. Gkizeli, D. A. Pados, and M. J. Medley, Optimal signature design for spread-spectrum steganography, IEEE Trans. Image Process., vol.16, no. 2, pp. 391 405, Feb. 2007 [8]. C. Fei, D. Kundur, and R. H. Kwong, Analysis and design of watermarking algorithms for improved resistance to compression, IEEE Trans. Image Process., vol. 13, no. 2, pp. 126 144, Feb. 2004. [9]C. Qiang and T. S. Huang, An additive approach to transform-domaininformation hiding and optimum detection structure, IEEE Trans. Multimedia, vol. 3, no. 3, pp. 273 284, Sep. 2001. [10] M. Gkizeli, D. A. Pados, S. N. Batalama, andm. J.Medley, Blind iterative recovery of spread-spectrum steganographic messages, in Proc. IEEE Int. Conf. Image Proc. (ICIP), Genova, Italy, Sep. 2005, vol. 2, pp. 11 14.