Improved performance of image watermarking by removing luminance bias in YCBCR channel

A digital watermarking for color image based on the modification of image pixels is considered in this paper. Its performance is analyzed and the weaknesses in the watermark embedding and extraction processes are addressed. Based on our analysis, two methods are proposed to improve the watermark embedding and extraction performance. First, the watermark signal is embedded in the chrominance channel. Second, the bias caused by the luminance component in the extraction process is removed. By applying our approach in the above watermark scheme, the impressive improvements in term of PSNR and NC are obtained, and verified by the experimental results. Its performance is also compared to the previous watermarking methods. We finally evaluate the robustness of our watermarking method against various common image processing based attacks.