Hybrid Motion Estimation Algorithm Based on PSO with Dynamic Threshold on Static Block Detection

Abstract

Motion Estimation(ME) technology is a platform in Video compression technology and so it is most complex process in video encoder. The challenge in the ME is to detect the best Motion Vector (MV) in less number of search points in the search plane. Pattern oriented search provides significant reduction on search points, especially the Hexagonal Search(HS) is much better, but last two decades the video codec’s demands acceleration on ME process to lighter encoder to adapt for even small scale hard cores. So there is necessary of optimization on search process. The proposed work optimize the ME process based on Particle Swarm Optimization PSO, and to improve performance of PSO its hybrid with HS. Also in this paper a method is proposed to detect the static blocks in image sequence by dynamic threshold. The PSNR and average search points per block are two metrics to measure the performance and its compared with popular search pattern techniques. The hybrid technique reduce the search points to 70% with the PSNR loss of 0.8dB. This method could be adapted to any video codec, to speed its ME process.