Spatioternporal sequence matching for efficient video copy detection

This paper proposes a novel sequence matching technique to detect copies of a video clip. If a video copy detection technique is to be effective, it needs to be robust to the many digitization and encoding processes that give rise to several distortions, including changes in brightness, color, frame format, as well as different blocky artifacts. Most of the video copy detection algorithms proposed so far focus mostly on coping with signal distortions introduced by different encoding parameters; however, these algorithms do not cope well with display format conversions. We propose a copy-detection scheme that is robust to the above-mentioned distortions and is also robust to display format conversions. To this end, each image frame is partitioned into 2 x 2 by intensity averaging, and the partitioned values are stored for indexing and matching. Our spatiotemporal approach combines spatial matching of ordinal signatures obtained from the partitions of each frame and temporal matching of temporal signatures from the temporal trails of the partitions. The proposed method has been extensively tested and the results show the proposed scheme is effective in detecting copies which have been subjected to wide range of modifications.