Motion-based parsing of compressed video

Abstract

An algorithm for detecting scene changes in compressed video streams is proposed. The proposed algorithm directly exploits the motion compensation information and the prediction error signal in the MPEG1-coded video stream. By performing minimal decoding of the compressed video stream, the proposed algorithm results in significant savings in terms of execution time and memory requirement. Experimental results show that the algorithm is effective in detecting abrupt scene changes (cuts) as well as gradual scene changes (dissolves) in an MPEG1-coded video stream. The algorithm is capable of processing video frames in real time and could be used for the rapid generation of key frames for real-time browsing of video streams and for indexing to support content-based access to on-line video libraries.