Research On Fast Retrieval Algorithms Of Video Fingerprint For Content Security

Within the rapid development of digital technology and network infrastructure construction, the speed of Chinese Internet has been further improved. The Internet video has became the main behavior of Internet users, and also became one of the major source of information. However, it seems to be hard to control the security of Internet video because of the growing number of video resource in the Internet. To solve this problem, we are inspired by the information hiding technology and biological sequence alignment technology, and developed a new way for the video searching.First of all, through detailed analysis of the current Internet video content security management problems and potential technology development at present situation, we argued that the digital fingerprinting has its unique advantages and feasibility in the field of video fast searching. We also put forward the problem that to be solved and the corresponding solution.Secondly, we analyzed and argued the principle of fast video searching system based on digital fingerprint that ensure the accuracy of video searching and improve the real-time performance. We also focused on researching the relevant technology and algorithm to improve the efficiency of video fingerprint searching.In order to continuously improve the efficiency of video fingerprint searching, making every effort to meet the needs of searching system for real-time, achieving high efficient and rapid searching of network video. This paper has proposed a fast searching mechanism based on the different thickness size searching. In the First step, this mechanism is using Location Sensitive Hash to hashing all the fingerprints in the database, and stored the resulting value in the hash table for further searching. Through this step, we can find the barrels that query sequence mostly resembled. This step has effectively reduce the video searching time. At the second step, in terms of detailed searching, we proposed the fast fingerprint matching mechanism based on biological sequence alignment technique of BLAST(Basic Local Alignment Search Tool). This mechanism not only are able to find the target video whether exist in the database quickly, but also find the video based on video clips. Even we can find the exact location that video clips in the complete video through this sequence alignment mechanism. At the last step of the mechanism, we using Clustal algorithm for video fingerprint to confirm the belonging of video based on the genetic traits and distinguish of video fingerprinting. These video that modified by the same original video will be seem as a family. This technology will convenient to manage video or multiple video searching.Thirdly, this paper designed a fast video searching system based on digital fingerprint. In this system, we verified and analyzed the accuracy and real-time searching performance of algorithm. And we summarized the main influence factors of the system.At last, we prospect the development of video searching technology and fingerprint technology, and summarized the defects of the algorithm in this paper.