Compressive Sensing for Image Watermarking Discrete Wavelet Transform and Spread Spectrum

Abstract

We propose a compressive sensing technique with LS Regularized Ll as the reconstruction method for image watermarking with some attacks applied to the system. Methods used in image watermarking are DWT and SS. The results of our experiment show that CS can be used as a compression method on image watermarking for some images due to the attacks which are decreasing the system performance. The best PSNR value can be reached is 52.35 dB and the best MSE value is 0.38 for host file D in the presence of JPEG compression attack. Meanwhile the best SSIM value is 0.67722 for host file C in the presence of JPEG compression attack. The best BER value is 0.26377 for host file E in the presence of JPEG compression attack. Our image watermarking system is more robust if JPEG compression attack is applied compared to other attacks simulated in this study.