| Nowadays,with the rapid development of Internet and multimedia technology, digital video based on video—coding has become more popular. There exists strong temporal correlation between adjacent frames in a video sequence. As an efficient tool for video compression, motion estimation (ME) and compensation (MC) can eliminate the high redundancy between frames and achieve a high compression ratio. Block-matching algorithm is simply and easily carried out by software and hardware, and is widely used in MPEG-1/2/4 and H.26x.The simplest and most reliable way is full search. Therefore, How to find an effective algorithm of motion estimation is one of the most popular topics in video-coding area.Motion estimation takes too much time in video-coding. In order to reduce the computational complexity and improve the accuracy of motion estimation, This paper analyzes the relationship between prediction motion vector and the optimal motion vector, and carry out a large number of statistical tests and simulation, concluded that the larger SAD value of prediction motion vector is, the more optimal motion vector distributed widely; On the contrary, the higher the concentration distribution. On this basis, we present a new block-matching algorithm in this paper: According to the prediction motion vector SAD value we set different thresholds and search templates. The smaller predicted SAD value is, the smaller the search range is. On the contrary, the wilder the search range is. Algorithm also uses the gradient-based descent search technology to improve the accuracy of the algorithm. Experimental results show that the algorithm is improved by 0.0016dB~0.0265dB in PSNR and decreased by 32.29%~ 45.32% in ME-time than UMHexagonS algorithm. |
PDF Full Text Request