Adaptive fuzzy search algorithm for improving the efficiency of block motion estimation

This paper proposes an adaptive fuzzy search (AFS) algorithm, based on the spatial and temporal correlation of image sequences, for improving the efficiency of block motion estimation. Due to the fuzzy reasoning method, the proposed algorithm could precisely predict the motion vector and reduce the computational cost. The essential difference between AFS and other related fuzzy methods is the adaptive ability for membership functions that are established according to the relationship of blocks and their neighbors. A Gaussian function is constructed through the differences between current block and its adjacent blocks. According to the Gaussian function of the current block, the weight of each adjacent block could be determined and then predict an initial center for the current block. This is why the AFS is adaptive and robust than other algorithms. Taking both the search point and block distortion measure into consideration, the performance of AFS is indeed better than the general inter-block/inter-frame fuzzy search (IIFS), TSS and LS.