Video fingerprinting in compressed domain: Fingerprinting method based on motion vector

In this thesis, a novel video fingerprinting method is proposed, in which the motion vectors extracted from the compressed stream are used to model the video fingerprint. Consequently, the generated fingerprint is then used to identify an unknown video, since a fingerprint should uniquely characterize a video. The identification process can be...

Variable bit rate compressed video

Fast Playback of Helical-scan Recorded MPEG Video