Research On Anti-Collusion Image Digital Fingerprinting Technology

With the rapid development of digital technology and popularity of Internet, digital technology brings us great convenience on the one hand, but makes piracy easier and consequences it caused increasingly serious on the other hand. As a new digital copyright protection technology, digital fingerprinting is a technology that embeds the unique identification user's information fingerprint into original digital products, which will be distributed to users. It functions as a copyright protection by extracting illegal fingerprint and using detection algorithm to realize piracy-tracking so that those pirates don't dare to apply his copies to piracy. Researches on digital fingerprinting technology are of great scientific significance and huge application values. The existing various fingerprint codes have a series of defects such as incomplete anti-collusion, long codeword, limited number of users, which have a long way to go. Practical digital fingerprinting application system has a lot of problems to deal with, especially the problem of anti-collusion.In this paper, we summarize several common fingerprint code schemes, analyze the merits and demerits of them, and introduce TWWL fingerprint scheme based on AND-ACC code and the linear combination collusion attack. To improve the robustness and enhance the anti-attack ability of fingerprint schemes, the main task has been done as follows:Hard detection arithmetic has been improved which sets an upper threshold for the correlator vector of traditional hard detection arithmetic to capture more colluders. The theoretical analysis and experiment results show that this arithmetic increased probability of tracking colluders.According to the characteristics of the TWWL scheme which is vulnerable to linear combination collusion attack, a digital fingerprinting scheme based on the image perceptive technology has been proposed. Fingerprints of the scheme are constructed in the form of TWWL's fingerprints in series. Embedding algorithm uses spreading spectrum embedding technology grounded on the spread spectrum of discrete cosine transform, and applies human visual model into arithmetic as well as utilizing improved hard detection arithmetic to track colluders. The scheme has been proved theoretically to be not only resistant to average attack, nonlinear collusion attacks such as maximum attack, minimum attack, but also enhance resisting linear combination collusion attack ability. At same time, the scheme has been proved practically to be effective by experiments.