Predictive Motion Estimation Structure for Motion Compensated Temporal Filtering
A. Kotegar Karunakar, M. M. Manohara Pai · 2005
The motion estimation algorithm consumes the dominant portion of encoding time in a video codec. A computationally efficient motion estimation technique is the need of motion compensated temporal filtering (MCTF) framework. In this paper, a novel predictive technique for fast block motion estimation is proposed in order to reduce the computational complexity by exploiting high correlation among the frames in a Group Of Picture (GOP). The proposed technique applies coarse and fine search, to the first pair of frames in a GOP and only fine search is applied around the predicted position for the remaining pairs. The technique is tested for different fast block motion estimation over the different standard test sequences. The results reveal substantial reduction (i.e. 41% to 66%) in the number of search points of motion estimation without compromising the quality of the received video compared to non-predictive techniques. Further, the cost of the motion vectors is reduced by 50%, for all the pairs except for the first pair in the GOP.