Research on Quality Evaluation Methods of Watermarked Printing Image Based on Image Dissimilarity

The watermarked printing image is a kind of multi-source heterogeneous image, which could be divided into the watermarking embedded process, the printing stage and the authentication phase. Image quality evaluation plays various roles in different image coding stages. In three stages, its functions are in turn providing direct modify data to optimize watermarking algorithms, evaluating robustness of the attacked watermark, and judging the veracity of product packaging, respectively, The idea of image dissimilarity expounds the presentation form, the characteristic direction and the generated reasons of differences before and after images distortion. It aims at using one-dimensional, two-dimensional or multi-dimensional spatial feature to express mathematically the macroscopic or microscopic transformation of the attacked image, At present, the classification method is uncommon that image quality evaluation is divided into image dissimilarity and image similarity. The image quality evaluation methods based on image similarity is numerous, while a few methods based on image dissimilarity are only presented from the two metrics of pixels gray value and singular values. But the latter could just makes up for its some shortages. Then according to the ideal of image similarity, it is feasible to in-depth research and establish the unified quality evaluation standard of watermarked printing image base image dissimilarity.