Adaptive edge masking based on TV decomposition and adjacent similarity for digital watermarking

Digital watermarking algorithms usually focus on texture masking due to the increasing noise insensitivity in texture regions. However, not all images have abundant texture regions. For images containing primarily edges and smooth regions, an image watermark should be embedded in edge areas. An adaptive edge-masking based on Total Variation (TV) decomposition for digital watermarking is proposed in this paper. First, edge detection based on the TV decomposition is applied to get accurate edges. After that, edge masking tends to be effective when the hidden signal has a similar direction and frequency as edges in the adjacent area. At last, the embedding intensity and position are determined adaptive to the image content. Experimental results show that our algorithm has good imperceptibility and is more robust than the algorithm without edge masking.