Published Paper Details:

Digital watermarking is a branch of information system security that finds tremendous applications in the modern era of internet and digital data exchange. Specifically, digital image watermarking finds vast applications in medical image processing, satellite images and related fields. In this paper, a blind and robust watermarking approach using DWT and SVD based transformations is presented. A given color image partitioned into RGB color planes and the entropy of each of the plane is computed for choosing the optimal plane for watermark embedding. The chosen plane is then segmented into non-overlapping blocks and watermarking bits are embedded into each of the blocks separately, considering the human visual system (HVS) characteristics based on the entropy values. For each of the block, the 2 level DWT is performed which transforms the image into four sub-bands namely HH, HL, LH and LL bands. Subsequently, Singular Value Decomposition (SVD) is performed on the HH band. The HH band is chosen to provide a robust watermarking as it contains most of the information contained in the original signal. Alternatively, LL band can be chosen to implement fragile watermarking scheme with a high PSNR values. To embed the watermark bits, the U and V matrix values are suitably modified to embed a zero or one. As compared to the S matrix of SVD transform, the embedding of the watermarking bits into the U and V singular matrices gives a much robust watermarking with low false positive rate and hence, are preferred when robustness is critical issue. The proposed technique is compared with those proposed by Nasrin M. et.al. and the simulation results established that better robustness and imperceptibility is achieved as indicated through PSNR.