Video Copy Detection Based On Video Fingerprint

Currently, multimedia technology are developing rapidly, new digital videos are generating all the time around the world. With the Internet, these rich video resources can be copied and disseminated in the case of almost no constraints. Efficient video content management and copyright protection is a huge problem. Video copy detection is important for video content management, which can be applied to video information retrieval, removal of the redundancy of video database, detecting and filtering specified videos with harmful content, commercial video tracking etc. Video copy detection is also an effective means of video copyright protection, which has great application requirements and prospects. Video copy detection should have characteristics of accuracy and real-time. Since the high distinction and robustness of video fingerprint ensures the accuracy of video copy detection, and the efficiency of video fingerprint search guarantees the real-time of video copy detection, this paper solves the problem of video copy detection based on video fingerprint, focuses on exploring methods of video copy detection that can meet both the requirements of accuracy and real-time .To meet the requirements of real-time video copy detection, the video fingerprint is generated based on the DC image sequences, the fingerprint is extracted only from the I frame, a large number of decoding time are removed, which improves the speed of copy detection. In order to ensure the accuracy of video copy detection, we improve the ordinal measure feature, limiting the impact of local change on the overall fingerprint, we also improve the motion feature, improving the accuracy of the motion feature, the video fingerprint generated by combining both of them has the advantage of higher accuracy. The experiments show, the proposed method of video copy detection is fast, compared with the current methods, the proposed video fingerprint has the advantage of higher discrimination and lower time complexity, the proposed method can meet the requirements of real-time video copy detection, and has comparatively strong robustness to the attacks of colour, size, gray, noise and so on.