高技术通讯2012，Vol.22Issue(1)：42-47,6.DOI:10.3772/j.issn.1002-0470.2012.01.007
基于压缩域的多路视频流混合方法
A DCT-based multipoint video composition scheme
摘要
Abstract
Aiming at the problems of high computational complexity and picture quality degradation of traditional pixel domain video mixing methods useful for multipoint conferences, this paper proposes a multipoint video composition scheme based on the compressed domain of discrete cosine transform (DCT) , and describes the details of the bits-tream mapping algorithm. With the rearrangement of the macroblock coding order and the mapping of the syntax element , the scheme combines multiple channel video frames together into the unique picture on the syntax layer according to the spatial position relation of the composition stream, and then a pre-quantization policy is particularly presented to remove requantization errors. To verify the availability, the details of the algorithm are integrated into the H. 263 codec. The experimental results revealed that compared with the cascaded method, the average peak signal to noise ratio of the proposed method (PSNR) was improved almost 2dB and the operational efficiency increased a hundredfold. It is possible that this research can provide a video mixing solution for the international video coding standard H. 265 which is under development.关键词
视频混合/压缩域/转码/视频会议Key words
video composition/ DCT domain/ transcoding/ video conference引用本文复制引用
王中元,胡瑞敏,常军,钟睿,韩镇..基于压缩域的多路视频流混合方法[J].高技术通讯,2012,22(1):42-47,6.基金项目
973计划(2009CB320906),国家重大科技专项(2010ZX03004-003-03)和国家自然科学基金(61003184,60970160, 61070080)资助项目. （2009CB320906）
