Enhanced Motion Estimation Algorithm based on Block Mean Deviation and Sorting Approaches

Abstract

Technologies such as TV, HDTV, 3D TV, Video telephony, Video surveillance, and Wireless multimedia communication have put a great demand for development of efficient and fast video compression algorithms. Motion Estimation (ME) is being used to achieve and increase the effectiveness of video compression. Traditional ME algorithms are not suited to current market demands. This paper proposes a Motion Estimation algorithm which is based on Partial Distortion Search and Sorting schemes. It tries to minimize the calculations by rejecting those partial distortion calculations early which would yield bad motion vectors. It is based on the concept that if deviation of sub-block is higher, then its contribution to total SAD (Sum of Absolute difference) is also higher. It uses this approach along with sorting of sub-blocks on the basis of their deviation. The results show that this algorithm can save 60 to 70 % of computation costs as incurred by traditional algorithms as well as performs well than other Partial Distortion Algorithms. Also PSNR degradation is very less and is 0.017 on average.