Multiresolution SVD and pixel-wise masking based image watermarking

Abstract

A multiresolution singular value decomposition (MR-SVD) based watermarking algorithm is presented here. The performance of this algorithm has the advantage of human visual system (HVS) based masking on the MR-SVD domain, which has been implemented pixel by pixel. The method employed here is adaptive adding of the watermark via texture and luminance content of the subbands obtained via MR-SVD. Correlation based detection of the watermark is used in this algorithm. The threshold of detection can obtained without the knowledge of the embedded watermark energy level. This feature enhances the adaptive feature of the algorithm for changes in host image. Experimental results provided here support for the effectiveness of the algorithm.