A Fast Hexagon-Based Search Algorithm on SIMD Architectures

Abstract

Due to the high computational complexity of the full-search algorithm in video coding, a hexagon-based search algorithm for block motion estimation has recently been proposed. In this paper, a fast hexagon-based search algorithm, suitable for implementation on single instruction multiple data (SIMD) architectures, is proposed to reduce the computational complexity of the original hexagon-based search algorithm without loss of its accuracy. Since the proposed partial sums are of only eight bits, byte-type data-parallelism on SIMD architectures can be utilized to further accelerate the execution of the proposed algorithm. Simulations are carried out for various benchmark video sequences and the results demonstrate that the proposed algorithm can reduce the computational complexity of the original hexagon-based search algorithm by 20 to 40 percent without loss of its accuracy. Simulation results also illustrate that the implementation on SIMD architectures can further accelerate the execution of the proposed algorithm by about 74 percent.