Research Of Content-Base Video Copy Detection Algorithm

Content-based video copy detection, referred to as CBCD, which is one of the important methods of digital video copyright protection and content management, widely accepted by domestic and foreign scholars.The paper mainly introduces two kinds of video copy detection algorithm, which are video copy detection algorithm based on sub-sequence and video copy detection algorithm based on key frame. Video copy detection algorithm based on sub-sequence utilizes the spatiotemporal information of video. First of all, We do the annular segmentation to each frame and get the gray scale of each block respectively, so that each frame get several annular segmentations from the inside to outside. Because the blocks at different positions allocate different weights, it embodies the spatial distribution of pixel and the different positions of the different importance of the blocks. Then we extract spatiotemporal information from the copy video and the original video, and calculate the corresponding spatial and temporal similarity. Then it makes use of corresponding threshold value to determine whether the video is a copy or not.However, the biggest drawback of video copy detection algorithm based on sub-sequence is that it has higher complexity of the algorithm, which brings inconvenience to the practical application. In order to achieve a breakthrough in that regard, the paper also introduces video copy detection algorithm based on key frame. This algorithm has video shot segmentation and key frame after annular segmentation. Then spatiotemporal similarity can be calculated by spatiotemporal information based on key frame. Both the dimension and the complexity of copy detection have been reduced rapidly.Finally, experiment show that copy detection algorithm is robust to rotation, added borders, added subtitles, image format and others geometric attacks.