2010 First International Conference on Integrated Intelligent Computing
Bit Plane Index Based Fragile Watermarking Scheme for Authenticating Color Image 1
Debjyoti Basu1, Arindam Sinharay2 and Suryasarathi Barat3 Bengal Engineering & Science University, Shibpur, 2 Future Institute of Engineering & Management, Sonarpur, 3 Director, PDSIT, Bengal Engineering & Science University, Shibpur 1
[email protected], 2
[email protected], 3
[email protected] Abstract
receiving to make sure that the received image is original or slightly modified. Digital image can be manipulated in many ways. Even in some cases, it is difficult for expert uses to discern whether an image is genuine. To address this issue, fragile watermarking has been developed for authentication and integrity verification of digital image. A commonly known fragile watermarking algorithm was proposed by Wong and Memon [1], where images are divided into over-lapping blocks of pixels and with block index. The resulting bitstreams are XOR-ed with binary logo, then encrypted and finally spread over the least significant bits of the pixels in each block. The detector must be acquainted with the indexes to extract the watermark from cover image. Therefore the detector can not estimate the originality of cover image. Based on the Wong and Memon’s work Celik et al [2] proposed a frame work that involves hierarchically structuring the pixels of the input image. The localization resolution achieved by this scheme might be deficient for some application. Fridrich [7] proposed framework where the authentication of data and the information about the origin of the image are separated. The resulting bit stream is XOR-ed with the fixed structure containing the image index, the location of block. The scheme uses private or public-key depending on encryption algorithm utilized. On the other hand Suthaharan [8] replaced the image index and the encryption algorithm with master key and session key used to generate a pattern by performing sequence of geometrical distortion into gradient image. In this method, the detector must be provided with exactly the same master and session keys as in embedding process in ordered to generate identical pattern. However each authenticator should be restrained to images watermarked with the same master key. In this paper, the current researchers propose a new fragile watermarking scheme that affords authentication even in the slightest modification of the watermarked image. Besides the propose framework embeds a 24-bit RGB image into the 24-
In the fragile digital color image watermarking for RGB color image authentication, fragility or sensitivity of the embedded watermark to malicious attacks is an important problem. In this work the current researches propose Bit Plane Index Modulation (BPIM) based fragile watermarking scheme for authenticating RGB color image. To deal with counterfeiting attacks block wise division dependency is established. A content based color watermark is created. Embedding distortion is minimized by adopting least significant bit (LSB) alteration scheme. The proposed scheme embeds watermark in three bit planes by changing original pixels with watermark pixels. The propose method comprise of encoding and decoding methods that can provide public detection capabilities in the absences of original host image and watermark image. The performances of the algorithm are verified by via experimental results and they illustrate the effectiveness of the proposed method. Keywords: Image authentication, counterfeiting attack, embedding distortion.
1. Introduction Digital multimedia has been frequently used for various applications due to easy transmission, coping, editing and storage. Subsequent, illegal distribution of and / or alteration of multimedia product are becoming more and more pervasive and digital watermarking techniques have been proposed [3] – [6] to solve such types of problems. Watermarking techniques can be divided into two broad types: robust watermarking and fragile/semifragile watermarking, which are used for different application purposes, from copyright protection to multimedia authentication. Fragile or semi-fragile watermark is commonly used for image authentication to verify whether the received image was modified during transmission or not. One may hide the watermark imperceptibly in the image before transmission and detect it after 978-0-7695-4152-5/10 $26.00 © 2010 IEEE DOI 10.1109/ICIIC.2010.53
136
bit RGB host image. This scheme is capable of detecting any modification done in any of the three bit plane individually, as explained in the following paragraphs. The rest of the paper is organized as follows. Section 2 and 4 explain embedding and extraction algorithms, section 3 and section 6 show the schematic diagrams of the respective algorithms, section 7 derives the conclusion and finally section 8 indicates the references.
3. Schematic Diagram of Embedding Process
2. Embedding Algorithm This algorithm actually embeds a color watermarking image into a color host image. The embedding algorithm is developed in such a way that it is perpetually impossible to identify the watermarking image. The schematic diagram of embedding algorithm is shown in figure 1. INPUT: Colour_Host_Image (Cij), Colour_Embedding_Image (Emn) Where, (i, j) ≥ 2 × (m, n) OUTPUT: Watermarked_Image (Wij) STEP 1: READ Cij, Emn ; STEP 2: Wij: = Cij ; STEP 3: Separation of Red, Green and Blue components of Wij ; Separation of Red, Green and Blue components of Emn ; STEP 4: Read Red Component of Wij (W_Rij); Read Green Component of Wij (W_Gij); Read Blue Component of Wij (W_Bij); Read Red Component of Emn (E_Rmn); Read Green Component of Emn (E_Gmn); Read Blue Component of Emn (E_Bmn);
Figure 1
4. Extraction Algorithm Proposed algorithm extracts the watermarking image from the watermarked image without the help of watermarking image and original host image. This is a kind of blind extraction method. After extraction it is found that the watermarking image is remained unaltered as the normalized correlation values of watermarking images are all unit. The schematic diagram is shown in figure 2.
STEP 5: For k: = W_Rij, W_Gij and W_Bij ; Begin Create four blocks (NWmn, NEmn, SEmn, SWmn); End STEP 6: For r : = E_Rmn, E_Gmn, E_Bmn Begin For s : = 1 to m Begin For t : = 1 to n Begin Read Binary value (8 bit stream) of each pixel; Create four groups of bits; Embed each group into W_Rij , W_Gij, W_Bij based on a mapping function; End End End STEP 7: Wij := W_Rij + W_Gij,+ W_Bij ;
INPUT: Watermarked_Image (Wij) OUTPUT: Extracted_Cover_Image (ECij), Extracted_Embedding_Image (EEmn) where, (i, j) ≥ 2 × (m, n) STEP 1: Read Wij ; STEP 2: Separation of Red, Green and Blue components of Wij ; STEP 3: Read Red Component of Wij (W_Rij); Read Green Component of Wij (W_Gij); Read Blue Component of Wij (W_Bij); STEP 4: For k: = W_Rij , W_Gij , W_Bij Begin Create four blocks (NWmn, NEmn, SEmn, SWmn);
137
original monochrome image and Kmn be the same image with noise. Then, Mean Square Error (MSE) can be calculated with the help of the following formula.
For l: = NWmn, NEmn, SEmn, SWmn Begin Read each pixel; Extract 2 LSBs; Replace 2 LSBs with ‘00’; End Create a bit stream from extracted LSBs, based on a mapping function; Write bit stream into an appropriate component of EEmn; Write replaced bit stream into an appropriate component of ECij; End STEP 5: EEmn = (EE_Rmn + EE_Gmn + EE_Bmn ); STEP 6: ECmn = (EC_Rmn + EC_Gmn + EC_Bmn);
From MSE, the PSNR can be calculated using the following formula
Where MAXI is the maximum possible pixel value of the image. If each pixel is represented by 8 bit stream then, MAXI = 255.
5. Schematic Diagram of Extraction Process
6.2 Correlation Coefficient Correlation Coefficient can be calculated using the following formula
where, A' = mean of the elements of the matrix A and B' = mean of the elements of the matrix B.
6.3 Experimental Results The table 1 below shows some of the experimented images with number of pixels Vs PSNR values of RGB components of images.
Figure 2
6. Experimental Data Analysis The current researchers use two types of parameter to analyze the effectiveness of the proposed set of algorithm.
Table 1 The table 2 below depicts the name of some of the experimental images, with their size and their PSNR values of RGB components.
6.1 Peak Signal to Noise Ratio (PSNR) It is a ratio between maximum possible power of a signal and power of corrupting noise. Let Imn be the
138
[2] M. U. Celik, C. Sharma, E. Saber, and A. M. Tekalp, “hierarchical watermarking for secure image authentication with localization,” IEEE Transaction on Image Processing, vol. 11, no. 6, pp. 585-595, 2002. [3] S. Katzenbeisser and F. A. E. Peticolus, Information hiding technique for Steganography and Digital Watermarking. Norwood MA: Artech House, 2000 [4] I.J. Cox, M. L. Miller, and J.A. Bloom, Digital Watermarking. San Diego, CA: Academic, 2002 [5] M.D. Swanson, M. Kobayashi, and A. H. Tewfik, “Multimedia data embedding and watermarking technologies,” Proc. IEEE vol. 86, no. 6, pp. 1064-1087, June 1998.
Table 2 The table 3 depicts the name of some of the test images and correlation coefficient values obtained from the comparison between the corresponding extracted embedding images and the original embedding image.
[6] F.A.P. Peticolus, R.J. Anderson, and M.G.Kuhn, “information hiding - a survey,” Proc. IEEE vol. 87, no. 7, pp. 1062-1078 July 1999 [7] J. Fredrick, “security of fragile authentication watermarks with localization,” in Proceedings of SPIE – The International Society for Optical Engineering, Ca, USA, 2002, vol. 4675, pp. 691-700. [8] S. Suthaharan, “Fragile image watermarking using a gradient image for improved localization and security,” Pattern Recognition Letters, vol. 25, no. 16, 1893-1903, 2004.
Table 3
7. Conclusion In this work, we have proposed a new scheme for authenticating 24-bit RGB images. By embedding R, G, B component of watermarking image in the R, G, B component of original image, the scheme is capable of detecting existing attacks. The balance between the security and embedding distortion can be adjusted by varying by the size of dependence on host and watermarking image in accordance with the need of the application.
8. References [1] P.W. Wong and M. Memon, “Secret public key image watermarking schemes for image authentication and ownership verification,” IEEE Transaction on Image Processing, vol. 10, no. 10, pp. 1593-1601, 2001.
139
