运动估计作为新一代视频压缩编解码标准H.264中的关键技术之一,因其巨大的计算量而成为了研究重点。该文针对非对称十字型多层次六边形格点搜索算法的不足之处提出了两个优化点,并引进了基于矢量分布的搜索模板以及螺旋子集匹配误差法,然后把优化后的算法集成到了H.264的参考模型JM10.1中。对优化后的算法进行实验后表明,在保证视频质量及编码速率几乎不变的条件下,编码耗时及运动耗时分别节省了10.683%和16.240%。
As one of the key techniques in the new video compression codec standards , motion estimation has become the concentration because of its tremendous computing workload . This paper proposes two aspects optimizations on the basis of the study of UMHexagonS , which include search pattern based on motion vector distribution and spiral subset matching strategy , and then integrates them into the reference model of H .264 named JM10.1.Experiment results show that the optimized UMHexagonS has encoding time and motion estimation reduced 10.683%and 16.240%on the condition of almost same video quality and transmission bit rate .