An adaptive watermarking algorithm based on edge detection

This paper presents an adaptive watermarking algorithm based on wavelet edge detection. The scheme compares the "edge and texture information after wavelet packet transforming" with the "sample through edge detection" of the host image to look for the most similar ones, which are quite right the most appropriate ones to embed watermark. And then computes embedded depth adoptively on the basis of texture characteristic of the selected block and its neighbors according to HVS, which can coordinate robustness and imperceptibility better. In the experiment, geometry cropping, Gaussian noise, mosaic etc. were applied as the attacks, and obtained a good achievement.