Photometric attack invariant video sequence matching

With the development of technology, volume of video data has grown heavily. It has become quite easy to capture, edit and copy such data leading to piracy. As a result copyright preservation has become quite crucial. A manual process of verifying a questioned video against a large repository is prohibitive and as a result content based video copy detection has emerged as an active area of research. To avoid copyright violation, a copier may incorporate different attacks on the copied data to give it a different look. Thus, an attack invariant detection scheme becomes essential. In this work we have relied on wavelet and edge based feature along with pre-processing steps to minimize the photometric attacks and finally, the hypothesis test based sequence matching adds to the strength in detecting copy in presence of the attacks under consideration.