Motion estimation with chessboard pattern prediction strategy

Due to high correlations among the adjacent blocks, several algorithms utilize movement information of spatially and temporally correlated neighboring blocks to adapt their search patterns to that information. In this paper, this information is used to define a dynamic search pattern. Each frame is divided into two sets, black and white blocks, like a chessboard pattern and a different search pattern, is defined for each set. The advantage of this definition is that the number of spatially neighboring blocks is increased for each current block and it leads to a better prediction for each block. Simulation results show that the proposed algorithm is closer to the Full-Search algorithm in terms of quality metrics such as PSNR than the other state-of-the-art algorithms while at the same time the average number of search points is less.