A Fast Motion Estimation Algorithm Based on the Adaptive Reference Frame and the Spatial and Temporal Correlations for H.264

In the H.264 video encoder, motion estimation (ME) is the most time-consuming component. The multiple reference frame search and the 7 partition modes block-matching are necessary in origin ME algorithm of H.264 and rate-distortion optimization (RDO) must be executed in each reference frame and mode in order to gain the best coding parameter. In this paper, a fast motion estimation algorithm based adaptive reference frame and the spatial and temporal correlations for H.264 is proposed. This algorithm well exploits the video sequence motion trend and the spatial and temporal correlations among the adjacent macro- blocks to decide and select the best reference frame and block matching mode quickly rather than searching in up to 16 reference frames and matching all size modes. The simulation experiment results show that the proposed algorithm save 38.4% ME time and loss only 0.018 db in PSNR than the origin H.264 algorithm with 5 reference frames.