Noise characteristics based postprocessing method in watermarking detection

Abstract

This paper proposed a postprocessing method to improve watermark correlating detection based on analysis of the characteristics of image distortion noise. The extracted watermark is classified and processed, separately, according to the characteristics of distortion noise. Analysis shows that the noise power can be reduced effectively. With the probability of false alarm kept unchanged, the probability of watermark detection can be increased. Experimental results demonstrate the effectiveness of the proposed method under different image distortions.