A framework for optimal adaptive DCT watermarks

In this paper we address the problem of robustly embedding 64 bits into an image while taking into account the H VS. The proposed method is general in that any mask can be adopted. The main advantage of the framework we present is that we demonstrate how to optimally embed a watermark given the constraints imposed by the mask in the spatial domain. This is in sharp contrast with the bulk of publications which embed a watermark in the DCT domain and then truncate or modulate in the spatial domain in order to satisfy masking constraints. The problem with these approaches is that spatial domain truncation or modulation leads inevitably to the degradation of the watermark in the DCT domain. Results indicate that our proposed approach is robust against JPEG compression at a quality factor of 30im-ages of size 64 by 64.