一种基于对象几何形状的视频抗误码方法

针对压缩视频流中的误码,提出了一种新颖的基于对象几何形状的视频抗误码方法.首先,在编码端通过比较帧间编码帧内邻近宏块的运动向量差异获得粗略的对象几何形状信息,以对象标示图表示;再把它以与标准码流兼容的方式嵌入到压缩码流里去;最后,在解码端利用它帮助精确的构造错误宏块的运动向量.试验结果表明,该方法对于即使大量连续slice出错的情况仍能获得比较好的掩盖效果,而增加的运算复杂度和传输负载量都不大.

An Error Resilient Method Based on Object Geometry for Video Streams

In this paper a new error resilient method based on object geometry is presented to deal with errors in compressed video sequences. First, object's geometry information is obtained at the encoder based on that the macroblocks belonging to the same object have similar motion vectors for an inter frame, and described as an object flag map. Then it is embedded into the video streams conforming to standard codec. And finally the object flag map is utilized to improve the accuracy of recovered motion vectors at decoder side. Experiment results show that our proposed method performs well even for a relatively large number of consecutive corrupted slices and the computational complexity and size of bit stream is increased limitedly.