Real Time Video Copy Detection under the Environments of Video Degradation and Editing

In the last decade, several effective algorithms have been developed in the areas of content based video retrieval and copy detection. However, the task of video copy detection becomes more difficult when the target video sequences available are in degraded or edited format. In this paper, we propose an improved method of video signatures generation and comparison to detect the copy of query video clip in the spatiotemporally altered real time target video streams. Furthermore, several experiments have been conducted with various sizes of rank matrices obtained from video frame partitions in order to infer the optimum sizes of the rank matrices.