International Journal of IT, Engineering and Applied Sciences Research (IJIEASR) ISSN: 239-443 Volume, No., October 202 8 Effect of Embedding Multiple Watermarks in Color Image against Cropping and Salt and Pepper Noise Attacks Balpreet Kaur, Deepak Aggarwal BBSBEC, FGS, CSE & IT Department ABSTRACT Multimedia technologies and internet have become part of our daily routine life. A wide range of digital data is created, transmitted, copied and distributed from one place to another through internet. So it is important to provide security and check for authorization of digital data. With images widely available on internet, it may sometimes desirable to use watermarks. A watermark is a secondary image, which is overlaid on the host image and provides a means of protecting the image. Watermarking is a very important field for copyright of various electronic documents and media. The research of this thesis aimed to present digital watermarking method for authorization against various geometric attacks. A binary watermark s embedded 6 times in diagonal manner in the host image, so there can be 6! (Factorial) combinations can be there to perform binary operations. The watermarks are embedded into different regions of blue component of the color image in spatial domain. The proposed scheme is invisible and blind. The watermark is a binary image, embedded into host image by altering LSB values of the selected regions. In this only 2 cases are considered for performing OR operations on extracted watermark, only one watermark will be selected based on highest NC value of extracted watermarks. In order to evaluate the performance of proposed algorithm, MSE (Mean Square Error), RMSE (Root MSE), PSNR (Peak Signal Noise ratio) parameters are used. The proposed scheme is found robust against various geometric attacks like cropping, and salt & pepper noise. Keywords RGB, watermark, PSNR, MSE, RMSE, NC.. INTRODUCTION A digital watermark is a digital signal or a pattern embedded into the host media to be protected, such as an image or audio or video. The digital watermarking technique is an effective way used for copyright protection and image authentication. It has become the research focus in international academics circles. There are two important properties of a watermark; the first is that the watermark embedding should not alter the quality and visually of the host image and it should be perceptually invisible, the second property is robustness with respect to image distortions. This means that the watermark is difficult for an attacker to remove and it should be also robust to common image processing and geometric operations, such as resizing, scaling, cropping, filtering and rotation.[] The process of watermarking involves the modification of original information data to embed watermark information. Various watermarking techniques have been developed. However these techniques are grouped into two classes: spatial domain and frequency domain. The spatial domain methods are to embed the watermark by directly modifying the pixel values of the original image LSB (Least Significant Bit) embedding is one of algorithm that uses spatial domain. [2]. The quality of watermarked image is measured by PSNR. Bigger is PSNR, better is quality of watermarked image. Watermarked Images with PSNR more than 28 are acceptable. Robustness is measure of immunity of watermark against attempts to remove or destroy it by image modification and manipulation like compression, filtering, rotation, scaling, collision attacks, resizing, copping etc. It is measured in terms of correlation factor. The correlation factor measures the similarity and difference between original watermark and extracted watermark. Its value is generally 0 to. Ideally it should be but the value 0.5 is acceptable. [2]. It is concluded that improved robustness against cropping attack can be achieved when multiple watermarks are embedded into different regions of host image. [8] 2. PROPOSED ALGORITHM The central idea of this thesis is to develop a watermarking algorithm that can embed more number of watermark bits and increase security and robustness of watermarks. In this spatial domain block based watermarking scheme is proposed. The watermark is a binary image and this watermark image will be embedded in the image 6 times in diagonal manner. Original digital image is a color image and 6 watermarks are embedded behind the image to achieve robustness against different attacks. A. Watermark Embedding The watermark image is a binary image and the host image is an 8 bit color image. The watermark is embedded 6 times in diagonal manner as shown in Figure in non
International Journal of IT, Engineering and Applied Sciences Research (IJIEASR) ISSN: 239-443 Volume, No., October 202 82 overlapping blocks of color image in order to protect the watermarked image.the 6 embedded positions are chosen to hide the watermarks in order to achieve robustness against cropping and noise attack in any order and intensity and make it difficult for attackers to destroy all of them. Suppose the original color image H with size of 52*52 pixels, which to be protected by the binary watermark W of size pixels 64*64. The original image H is divided into five6 non overlapping blocks and size of block equals to the size of watermark image i.e.64*64. Algorithm: Input: Color (original) Image (C) and binary Watermark image (W). Step : The original image C is taken as input. Now from this image R, G and B components will be separated. The image with blue components will be divided into6 non-overlapping blocks. Size of each block equals to the size of watermark image i.e. 64*64. Step 2: For embedding the watermark in the blue component image, the intensities of blue component image are converted into binary. Binary watermark is embedded 6 times diagonally into the LSB of blue component image. Because the watermark is binary it includes either 0 or which is added into binary value of LSB. Step 3: After embedding 6 watermarks diagonally behind blue component image, the original color image will be obtained by adding red component image, green component image and blue component image along with watermarks. Step 4: After getting the final color image with embedded watermarks the original image and the image with embedded watermarks will be compared by taking into consideration different parameters. B. Watermark Extraction: Watermark extraction does not required the original host image and the original watermark; therefore, it is a non blind watermarking scheme. The steps of watermark extraction are: Step : 6 watermarks (w- w6) will be extracted from 6 positions in diagonal manner by extracting from LSB of pixel intensities of color image. Fig. Watermark embedding in diagonal manner Step 2: After getting original watermark we can perform OR operation b/w different watermarks. There are 6 watermarks in diagonal manner so there can be 6! (Factorial) combination to recover the watermark by performing OR operations but here only 2 cases are generated for performing OR operations. The cases are: Case : w OR w2 Case 2: w OR w9 Case 3: w OR w6 Case 4: w8 OR w9 Case 5: w8 OR w6 Case 6: w OR w2 OR w3 Case : w OR w3 OR w5 OR w Case 8: w9 OR w OR w OR w5 Case 9: w2 OR w4 OR w6 OR w8 Case 0: w0 OR w2 OR w4 OR w6 Case : w OR w9 OR w8 OR w6 Case 2: w4 OR w2 OR w5 OR w3 Step 3: After applying all the above 2 cases of OR operations the watermark will be extracted from the image. Step 4: Now we calculate the NC (normalized correlation) extracted watermark through OR operation with the original watermark to check the similarity between original and extracted watermark. Now the OR operation of watermarks with the highest NC is considered as the final watermark. Step 5: After this various geometric attacks like cropping and salt and noise will be applied on the image to check the robustness of the watermark.
International Journal of IT, Engineering and Applied Sciences Research (IJIEASR) ISSN: 239-443 Volume, No., October 202 83 3. EXPERIMENTAL RESULTS To verify the effectiveness of the proposed method, a number of experiments are conducted. In the experiment, MSE=σ 2 2 q = Where the sum over j, k denotes the sum over all pixels in the image and N is the total number of pixels in an image. The lower the value of MSE, the lower the error. RMSE= A lower value for RMSE means lesser error and this result in a high value of PSNR. PSNR= 20 * log0 (255 /RMSE) Where n is the number of bits used to represent per pixel value and 255 represents the maximum value of each pixel. Logically, a higher value of PSNR is good because it means that the ratio of signal to noise is higher. So we can say that a scheme having a lower RMSE and a high PSNR is a better scheme. an original image is color image of size 52x52, where as the watermark image is a binary image of size 64x64 pixels. To evaluate the efficiency of algorithm MSE, RMSE and PSNR are used, which are defined as follows extracted and even its quality is not degraded. When the image is cropped 25% from the top or from the bottom even then the results are efficient. Fig 3. 50% cropping Fig 4. 5% cropping When the image is cropped to 95% and 98% even then the watermarks are extracted and the output is efficient. In almost all the cases the value of PSNR is INF, MSE is 0, RMSE is 0 and value of NC is. Fig2. Original Image To test the robustness of the proposed scheme, some attacks such as salt and pepper noise, cropping is performed. 4. EFFECT OF ATTACKS: The performance of proposed algorithm is analyzed by considering image processing attacks like cropping and salt and pepper noise. A. Cropping The watermarked image is cropped in terms of percentage of image size like 25%, 50%, one quarter, 5%, 95% and 98%. It can be seen clearly that the watermark can be extracted correctly under various cropping attack. In case of 50% cropping of the whole image the watermarks are extracted and even its quality is not degraded. When the image is cropped 25% from the top or from the bottom even then the results are efficient. It can be seen clearly that the watermark can be extracted correctly under various cropping attack. In case of 50% cropping of the whole image the watermarks are Fig. 3. 98% cropping Table. Cropping under different %age CROPPI PSN MSE RMSE NC NG R 25% Inf 0 0 50% Inf 0 0 5% Inf 0 0 One Inf 0 0 quarter 95% 66.54 0.0440 0.200 59 43 8 98% 66.4 29 0.0464 84 0.20 3 0.98 338
International Journal of IT, Engineering and Applied Sciences Research (IJIEASR) ISSN: 239-443 Volume, No., October 202 84 In table the result of attack cropping is shown. As it is shown that till one quarter cropping the value of PSNR is INF, MSE is 0, RMSE is 0 and NC is. In other two cases the results are efficient. They have better PSNR, MSE, RMSE and NC. B. SALT and PEPPER NOISE The salt and pepper noise is added to the watermarked image. The performance of extraction algorithm is analyzed by increasing density of the noise starting from 0.0 to.0. The extracted watermark and original watermark are compared in terms of NC. In table 2 all cases with density 0.0 are shown. From table it is cleared that in all the cases the value of NC is. Even it shows that all the cases have better PSNR, MSE and RMSE values. Because the value of NC is so it is cleared that the watermark is completely extracted. Table2. 2 cases of or operation under noise density 0.0 CASE Case : w OR w2 Case 2: w OR w9 Case 3: w OR w6 Case 4: w8 OR w9 Case 5: w8 OR w6 Case 6: w OR w2 OR w3 Case : w OR w3 OR w5 OR w Case 8: w9 OR w OR w OR w5 Case 9: w2 OR w4 OR w6 OR w8 Case 0: w0 OR w2 OR w4 OR Case : w OR w9 OR w8 OR Case 2: w4 OR w2 OR w5 OR PS NR 9. 4832 4. 2 2646 9. 4832 264 5. 8034 8405. 9499 8405 5 2. 4935 264 MSE RMSE N C 0.00032 422 0.0029 2 0.00220 0.00032 422 0.00220 0.0008 98 0.002685 55 0.00450 39 0.002685 5 0.0033 83 0.003662 0.00220 0.020 633 0.0468 5 0.3493 86 0.020 63 0.0349 386 0.043 399 0.058 223 0.0644 235 0.058 223 0.0563 36 0.0605 54 0.0349 386 From fig 4 it is clear that the proposed scheme has good PSNR value and the value of MSE is near to 0 which is very low. Fig. 4 Effect of salt and pepper noise 5. CONCLUSION A robust watermark scheme based on a block base for color image is presented which operates in spatial domain by embedding the watermark image 6 times in diagonal manner in order to achieve robustness against various geometric attacks like rotation, cropping, salt and pepper noise and speckle noise. The original image is not needed in the detection process, so it is a non-blind watermarking. The experimental result shows that this scheme is highly robust against various image processing operations such as cropping with different %age and salt and pepper noise from 0.0 to.0. REFERENCES [] Alankrita Aggarwal, Monika Singla Robust Watermarking of color Images under Noise and Cropping Attacks in Spatial Domain, International journal of computer science and Information Technologies, vol, 2(5), 2036-204, 20. [2] Baisa L. Gunjal and R.R. Manthalkar, An overview of the transform domain robust digital image watermarking algorithms, journal of Emerging Trends in Computing and Information Sciences, Vol. 2, No., ISSN 209-840, 200-20. [3] B Surekha and Dr. GN Swamy, A Spatial Domain Public Image Watermarking, International Journal of Security and its Applications, Vol. 5, No., January, 20 [4] Darshana Mistry Comparison of digital watermarking Methods (IJCSE) International Journal on Computer Scienceand Engineering vol. 02, No. 09,, 2905-2909,200. [5] Heather Wood, Invisible digital watermarking in the Spatial and DCT Domain for Color Images.
International Journal of IT, Engineering and Applied Sciences Research (IJIEASR) ISSN: 239-443 Volume, No., October 202 85 [6] Ibrahim Nasir, Jianmin Jiang and Ying weng, school of computing, Informatics and Media, University of Bradford, Bradford BD DP, U.K Novel Multiple Spatial Watermarking Technique in Color Images, IEEE, 2008 [] Ibrahim Nasir, Ying Weng, Jianmin Jiang, School of Informatics, University of Bradford, U.K A New Robust Watermarking Scheme for Color Image in Spatial Doman, IEEE, 2008 [8] Ibrahim Nasir, Ying Weng, Jianmin Jiang, Stanley Ipson, School of Informatics, University of Bradford, U.K Multiple Spatial watermarking techniques in color images, Springer, 200. [9] K. Ganesan and Tarun Kumar Guptha, Multiple Binary Images Watermarking in Spatial and Frequency Domains Signal and Image Processing: An International Journal (SIPIJ), Vol.. No. 2, December 200. [0] Keshav S Rawat, Dheerendra S Tomar, Member IEEE, Digital watermarking schemes for authorization against copying or piracy of color images, Indian Journal of Computer science and Engineering, vol. No. 4 295-300. [] Mansi Hasija, Alka Jindal, Contrast of watermarking techniques in different Domains, IJCSI, international Journal of Computer Science Issues, Vol. 8, Issue, No. 2, May 20. [2] Nagraj V. Dharwadkar and B. B Amberker Secure Watermarking Scheme for Color Image Using Intensity of Pixel and LSB Substitution journal of computing vol., Issue., ISSN: 25-96.December, 2009, [3] Ramani K.; Prasad, S Varadarajan, S.; Subramanyam A, A Robust Watermarking scheme for information hiding advanced Computing and Communications, 6 th International Conference, pp: 58-64 4-, Dec. 2008.