针对样品视频的H.264/SVC的快速运动估计算法

针对应用于电子显微镜等大型科学仪器样品图像远程实时传输的H.264/SVC,提出了一种基于自适应搜索范围的快速运动估计算法。根据宏块的空间和层间相关性,将基本层及增强层宏块分类,以宏块运动向量残差为基础,针对不同类型宏块,在水平和垂直方向上设计了自适应搜索范围算法。并针对匹配块搜索过程和帧间预测过程分别设计了提前退出算法。对不同仪器样品视频的测试表明,该算法与参考软件JSVM9.17的快速搜索算法相比,在PSNR和比特率保持相近的情况下,编码时间大约可节省52%～68%,搜索点数大约可降低81%～95%。

H.264/SVC fast motion estimation algorithm for sample video

A fast motion estimation algorithm based on adaptive search range is proposed for the H. 264/SVC,applied in remote real-time transmission of sample images of electron microscopes and other scientific instruments.The macroblocks at the base layer and enhancement layer are classified according to the spatial and inter-layer correlation.For the macroblocks of different types,an adaptive search range algorithm in both horizontal and vertical directions is designed based on macroblock motion vector difference.In addition,early termination algorithms are also designed for matching block searching and inter-frame prediction respectively.Experimental results for sample videos of different instruments show that,compared with the fast search algorithm of reference software JSVM9.17,the proposed algorithm can reduce the encoding time by 52%-68%and the search points by 81%-95%,with similar PSNR and bit rate.