Symmetrical-cross Multi-Hexagon based Exhaustive Search for block motion estimation

In this paper, we propose a new efficient algorithm for block motion estimation, called Symmetrical-crossMulti-Hexagon based Exhaustive Search (SMHES). It consists of two steps: 1) Motion Prediction and 2) Quadrant Search. First step uses symmetrical-cross, small square and multi-hexagon-grid search over the entire search range to predict the direction of motion. Second step involves exhaustive search of the quadrant in which predicted motion (minimum SAD point of first step) occurs. Simulation results show that this algorithm achieved 70% reduction in number of computations with negligible loss (<0.02dB) in PSNR when compared to Full Search Algorithm. Our implementations used standard test sequences in SIF and CIF formats which are supported by low power mobile device video applications.