PSO Optimized Multipurpose Image Watermarking Using SVD and Chaotic Sequence

This study proposes a novel method for multipurpose image watermarking for both ownership verification and tampered region localization. Two watermarks (robust and fragile) are inserted into the host image. Robust watermark insertion is done by PSO (particle swarm optimization) optimized scaling of the singular values; utilizing the singular value decomposition (SVD). Doing so, leads to reduction in visibility changes (better imperceptibility) of host image as well as enhanced performance of watermarked image towards attacks (better robustness). Fragile watermark insertion is done by making use of SVD and chaotic sequence (block feature's dependent). The image is first divided into non overlapped blocks and block based Arnold transformed is performed. Then after, block grouping is done of scrambled blocks to breakdown their independence in order to sustain the vector quantization and collage attacks. The proposed scheme is tested against various signal processing attacks and results shows a good performance.