Efficient Motion Vector Coding for IBMCTF Framework
H Anitha, M. M. Manohara Pai, A. Kotegar Karunakar · 2003
A new In-Band Motion Compensated Temporal Filtering (IBMCTF) framework is introduced to provide scalability for compressed bitstreams of video. The number of motion vectors generated for this framework is too high. The existing prediction technique to compress the motion vectors has three different modes for prediction, which is to be transmitted to decoder and hence requires minimum two bits. In this paper we are proposing only two modes instead of three (i.e. one bit instead of two). Thus the cost of mode information to be transmitted to the decoder is reduced by 50%, without compromising the quality of the received video.