A Robust and Oblivious Grayscale Image Watermarking Scheme Based on Edge Detection, SVD, and GA

For multimedia data copyright protection application, robustness to various attacks is one of the most important requirements that a digital watermarking system should possess. Hence, using matrix factorization technique (i.e., singular value decomposition) and genetic algorithm, a new grayscale image watermarking system is presented in this paper which can satisfy the desired watermarking requirements. Canny edge detector is used to form two subimages and watermark is inserted into the first subimage by amending singular value coefficient based on the pixel value of a watermark. The genetic algorithm is employed in the proposed scheme to search the best multiple scaling factors which can give the highest robust watermarked image without losing transparency. The robustness and transparency of the scheme are measured using a quality metric, normalized-correlation-coefficient (NCC), and peak-signal-to-noise-ratio (PSNR), respectively. In this paper, to test the degree of robustness, watermarked image is attacked using a maximum number of image processing attacks compared to other existing methods and experimental results obtained show improved performance in terms of robustness.