An improved block matching motion estimation algorithm based on adaptive rood pattern search

Abstract

Block matching motion estimation is a significant issue in image and video super resolution reconstruction. In order to decrease the number of search locations, the amount of computations, and enhance the prediction accuracy, an improved block matching motion estimation algorithm based on adaptive rood pattern search (ARPS) is proposed. Firstly, we use spatial correlation characteristic to predict motion vector (MV) of the current macro-block in the algorithm. Secondly, we consider to eliminate the influence of neighborhood macro-blocks having low relevance to the current macro-block in the prediction process. Finally, we utilize the weighted average MV of two macro-blocks with the smaller matching error in the neighborhood macro-blocks to calculate the predictive MV of current macro-block, instead of using the fixed single macro-block forecast pattern in ARPS. Experimental results show that our algorithm provides the better performance in terms of the computational complexity and the prediction precision in comparison with APRS and other block matching algorithms.