A new way to reduce candidate blocks for block matching motion estimation

A new way to reduce candidate blocks for the block matching algorithm is proposed. It consists of three steps: (1) compute the average luminance of all the concerned blocks by a fast iterative algorithm; (2) calculate the absolute error between the average luminance of the currently concerned block and that of the candidate block within the search window in the previous frame; and (3) remove those candidate blocks with large difference. Then the most similar block will be found by searching the remaining candidate blocks. The complexity of reducing the search space is very small, but it can remove candidate blocks from the search window efficiently and effectively. The simulation results are given to demonstrate the effectiveness of this new algorithm