An Adaptive Color Image Watermarking Scheme Using Combination of PSO-LTSVR in Wavelet Domain for Copyright Protection

Abstract

AbstractWith the advancement of internet technology, copyright protection and ownership assertion of multimedia content (audio, images and videos) is becoming a more challenging task. During last two decades various researchers raised the issue of watermarking for copyright protection and ownership assertion of multimedia contents, but still trying to find the better solution than the existing approaches. In order to handle the issues of imperceptibility and robustness evolved in color image watermarking, a novel color image watermarking approach in YCbCr space using discrete wavelet transform (DWT) and combination of particle swarm optimization (PSO) and Lagrangian twin support vector regression (LTSVR) is proposed. The low frequency wavelet coefficients of selected regions based on the human visual characteristics (HVS) is selected to embed the watermark. Considering the watermarking problem as an optimization and regression problem, PSO is used to find the optimal value of watermark strength to embed the watermark and LTSVR is used to extract the watermark against image processing attacks. With the combination of PSO-LTSVR in wavelet domain, the proposed approach reduces the trade-off between imperceptibility and robustness. The experimental results tested on different textured color images clearly reveal the imperceptibility of the watermarked images and robustness of the proposed scheme against JPEG compression attacks with different quality factors (QF) as quantified by peak-signal-to-noise ratio (PSNR) and bit correct ratio (BCR).KeywordsDWTPSOLTSVRPSNRBCR