The Study On Fast Motion Estimation Algorithm Based On AVS

In video coding and processing system, motion estimation (ME) plays very important role in eliminating inter-frame redundancy and improving the performance of video coder. On one hand, the accuracy of motion estimation affects the efficiency of the video coder. On the other hand, the complexity of the encoder lies on that of the of motion estimation algorithm. The motion estimation and motion compensation techniques based on AVS is mainly discussed in this paper.By analyzing the motion characteristic of video sequence and the advantages/disadvantages of the classical searching algorithms, a new fast searching algorithm is proposed based on prediction of motion vector and self-adaptive searching area. In this algorithm, the motion vector is predicted by the space/time correlation of the motion vector. The predicted value of current block is equal to mean value of 4 referential motion vectors, and the starting point of the searching algorithm is chosen as it. At one time, the searching area is decided based on the similar class of the reference motion vectors. Experimental results prove that the searching speed of the new algorithm increases 75 times compared with Full Searching (FS) algorithm, 4 times compared with Diamond Searching (DS) algorithm. The PSNR is about 0.35db lower than FS and about 0.5db higher than DS. The new algorithm is suitable for real-time applications because it achieves similar performance compared with FS and much faster than DS.