Digital watermarking for color video using a nonlinear filter in detection process

A new watermarking method for color video using a nonlinear digital filter is proposed. This method is based on Haitsma and Zhao´s methods, which embed watermark signals into the luminance mean series, but introduces a new nonlinear digital filter, named component separating filter (CS filter), in the detecting process. The CS filter can effectively extract the small-amplitude watermark signals on the mean series without being affected by abrupt shot changes. Moreover, it is shown that higher performance is realized by embedding into the color difference signal instead of the luminance. Computer simulations verify its high performance compared with conventional methods.