H.263视频场景自适应编码的改进算法  

Improved scene adaptive encoding algorithm for H.263 video stream

作　　者：李虓江[1] 陈抗生[1] 

机构地区：[1]浙江大学信息与电子工程系,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2004年第3期292-296,共5页Journal of Zhejiang University：Engineering Science

摘　　要：针对现有场景自适应编码方案中存在的检测结果不准确问题,引入图像相似度检测算子,提出了一种基于二级筛选机制的图像相似度检测方案.在使用直方图法进行图像相似度一级检测的基础上,利用相似度检测算子对检测结果进行二级筛选,剔除一级检测中引入的伪突变帧,找出真正的突变帧.然后对该突变帧进行帧内编码,开始一个新的视频片段,从而使由于场景切换所造成的视频图像失真得到有效抑制.对标准视频序列的仿真试验表明,提出的改进方案可以更准确地检测出序列中的突变帧,从而降低输出编码视频的比特数,提高视频压缩效率.H.263 recommendation can provide effective compression method under very low bit rates. However the visual quality of some encoded video segments declines when the scene changes in the video. A new scheme based on two layers filtration mechanism was proposed. In this scheme, an image similarity examination operator reexamined the result of histogram examination algorithm in order to get rid of the fake abrupt frame. After filtration operation, the true abrupt frame was coded as intra-frame and then a new video segment began. Simulation results show that the proposed scheme can effectively restrain the video distortion during scene changes. The abrupt frame will be filtered out more exactly and the bit-rate of encoded video is less than that of former schemes.

关 键 词：H.263建议 场景自适应编码 二级筛选机制 图像相似度检测算子 

分 类 号：TN919.8[电子电信—通信与信息系统]