Abstract:

In order to improve the performance of spatial image watermarking algorithm, a kind of prediction error filter based on cover image Least-Square (LS) was proposed, which is applied to the embedding and detecting processes. In the embedding process, the filter was used to produce a new spatial perceptual mask. The cover image prediction error sequence controls the watermark strength and matches well with the properties of human visual system. In the same visibility quality metrics, the scheme can embed more watermark strength. In the detection process, in order to reduce the cover image effect on the correlation detection, the LS prediction error filter was applied to the received image and estimation of the masked watermark. The normalized correlation measure for the resulting error sequence was computed. It is proved that the scheme is robust to general signal processing and geometric attacks.

Key words: information processing, digital watermarking, masking function, prediction error filter