基于改进KNN算法的AVS到H.264/AVC快速转码方法  

Fast Transcoding From AVS to H.264/AVC Based on Improved KNN

作　　者：沈皓[1,2] 

机构地区：[1]上海交通大学,上海200240 [2]中国南方航空上海分公司,上海201202

出　　处：《电视技术》2015年第8期35-39,共5页Video Engineering

摘　　要：尽管音视频编码标准(Audio and Video Coding Standdard,AVS)的编码性能可以与H.264相媲美,但是H.264的应用范围更加广泛,因此视频由AVS标准转码成H.264标准具有很大的应用前景。目前,主流的转码方法是将AVS的分块模式与H.264的分块模式映射的方式降低转码复杂度,但是技术之间的差异导致这两种标准之间的分块模式并不是一一映射的关系,因此会导致编码效率大幅度降低。提出一种基于改进KNN(K最邻近节点)算法的AVS到H.264/AVC快速转码方法。充分利用了AVS码流中的各种信息,通过改进的KNN算法建立了中间信息和H.264分块模式之间的映射模型。根据AVS中运动矢量信息的差异自适应确定H.264可能的分块模式,实验结果表明上述问题得到有效解决,该算法在保证H.264编码效率的前提下大幅降低了转码复杂度。Although the encoding performance of Audio and Video Coding Standard （AVS） can be comparable to H.264/AVC, H.264/AVC has more extensive applications. So the transcoding from AVS to H.264/AVC has great application prospects. Recently, the mainstream transeoding methods reduce the complexity according to the mode mapping method between AVS and H.264/AVC. But for the different techniques between the two standards, the one-to-one mapping of modes is impossible, thus causing a significant reduction in the encoding efficiency. A fast mapping algorithm based on improved KNN is proposed in this paper, which makes full use of various information of AVS bitstream. And the possible mapping mode can be determined adaptively when utilize the motion vector information of AVS. Experimental results show that the above issues are effectively addressed. Large numbers of experimental results show that the proposed method significantly reduces the transcoding complexity under the permission of transeoding efficiency decrease.

关 键 词：音视频编码标准 快速转码 K最邻近结点算法 

分 类 号：TN919.81[电子电信—通信与信息系统]