An Improved Adaptive Cross Pattern Search (IACPS) Algorithm for Block Motion Estimation in Video Compression
Amit Chatterjee, Rana Mukherji, Sudhanshu Tripathi · 2007
Motion estimation is a key issue in the field of moving images analysis. Since motion estimation is the most time taking part in the video compression, nearly it takes the 80% of the total time. And the timing efficiency of the motion estimation mainly depends upon the fact, that how we are searching the motion between the two successive frames of video sequences. So, it depends upon the total number of search points for finding the motion vector in the search window. If we are searching less number of points, maintaining the quality of the video then it will lead to less computation and it will improve the speed of motion estimation. The proposed IACPS algorithm utilizes a predictive motion vector and a cross search pattern. Simulation results show that the proposed IACPS algorithm provides similar PSNR performance for large and medium motion sequences but worst in the case of only camera movement. And computation per macro-block is better than all algorithms, except ARPS in all the type of motions. But IACPS provides best computation per macro- block in case of video sequences with medium motion and low spatial detail.