An analysis of independence of video signatures based on tomography

The current technological advances allow for the easy distribution and duplication of video content through various venues, the most popular being the World Wide Web. Upholding copyright laws has thus become an important task; the result to this problem needs to be implemented in such a way to provide automatic video identification. This paper presents a novel approach to video identification based on the video tomography technique. The proposed video signature was designed and evaluated based on its ability to uniquely identify videos. A video signature should be both independent and robust. This paper focuses on evaluating the independence of the proposed video signatures. The signatures were evaluated using a large test set developed for the MPEG activity on digital video signatures. The results show that the proposed signature works well and exhibits strong independence necessary to support general purpose video identification systems.