The Research On Collusion-secure Fingerprinting System

With the rapid development of Internet technology, it is becoming more and more convenient to exchange the multimedia data. Subsequent various piracy is becoming a serious threat for protecting the data. As an effective copyright protection technology, digital fingerprint is becoming more and more important. Digital fingerprint technology protects the digital products by embedding some of the personal information and (or) purchase information, and store in the database of vendor. When publishers find the pirated copy, they can extract the corresponding fingerprint information, and comparison with fingerprint information in the database, judging the illegal distributor, and find who is to blame the law responsibility. To design collusion-secure digital fingerprint scheme is very important.In this paper, we first introduce related concepts, such as models, attack, embedded and extract methods of digital fingerprint. Then we introduced two classical discrete coding schemes. The main contribution of the thesis is as follows:Firstly, according to the advantages of orthogonal fingerprinting and the principles of connected orthogonal code, we present a new fingerprinting scheme based the orthogonal fingerprinting and LDPC codes. The theoretical analysis shows that this fingerprinting scheme can resistant the large number of colluders and has excellent performance in error correction. We also give the embedding and extracting model under this fingerprinting scheme.Secondly, we improved maximum detection algorithm and threshold detection algorithm on the basis of analysising the detection scheme of orthogonal fingerprinting and math expression formula of max-colluders under theoretical analysis standard of capturing the max-colluders and all colluders. The original detection in the process has the certain error for the reason that the number of colluders in the detection of maximum detection algorithm and threshold detection algorithm is unknown, and in the improved detection, we first estimate the max-number of colluders, and then, determine the threshold under designated by mistake examining rate. Through this method, we reduce the certain error and improve the detection efficiency. The simulation experiment shows the improved orthogonal fingerprint detection method for detection efficiency performance have been greatly improved than the original program.