Study On Anti-Collusion Digital Fingerprint And Embedding Algorithm Under New Image Quality Index

As the development of the Internet and the improvement of the degree of information-exchange of the whole society, an increasing number of image and video are casting an influence on our lives in a digital form. These in the meantime set a platform for those who copy illegally and redistribute the work. Because traditional cryptography will lose the ability to trace the enemy after decryption, digital watermark system occurred and solved this problem in some degree. But normal watermark has its own limits, and there comes the digital fingerprint. However, even we have used digital fingerprint, there is no certainty that the fingerprint can not be damaged under current anti-attack technologies. If couples of users make a coalition and collude to generate a new image or video section, they would weaken the examinable energy of the fingerprint a lot, hence avoid being traced. Colluders can even let the fingerprint appeared to be very similar to some innocent user, thus frame other people in order to escape from taking responsibilities. So far the study on anti-collusion fingerprint can be widely divided into two main directions, one is designing anti-collusion code, and the other is analyzing the performance of anti-collusion system. This thesis summarizes some cross-sectional results related to the two and some other aspects, and brings out one of the main difficulties in the research of anti-collusion algorithm—code length problem.Usually, because resisting the collusion attack needs more information to be embedded in the original work, for anti-collusion fingerprint, whether the embedding capacity is adequate on demand or not is a key factor. Ground on the comparison between the former image quality index based on error visibility and the new image quality index based on structural similarity, the thesis gives an examination on the problem of the embedding capacity of fingerprint information standing on the new index function's angle,and proposes an image-adaptive embedding algorithm in DCT domain based on this image quality index. Experiment result shows that, if the total embedding capacity is relatively small: when embedding with the same capacity, the visual quality of the new embedding algorithm is apparently better than average embedding method; while maintaining almost equivalent perceptual quality, the embedding capacity of the new algorithm is rather higher than the average one.The thesis gives some analysis of the new embedding algorithm, and compares the new algorithm with other embedding algorithms, at the last part reviews all the work of the thesis systematically, and proposes some feasible directions on the study of anti-collusion algorithm and the prospect of the new embedding algorithm.