用于多视视频加深度的错误隐藏算法

An error concealment method for multi-view video plus depth

提出了一种针对多视视频加深度(MVD)的错误隐藏(EC)算法。算法充分利用MVD特有的深度信息及当前丢失宏块周围正确解码的宏块信息,将丢失宏块分为3类不同属性的宏块。针对3类宏块的特点,分别提出了基于候选运动矢量修正(CMVR)、基于深度的外边界匹配(DOBMA)以及自适应权值的EC(AWEC)等模式。实验表明,本文提出的算法在保证相同的视频主客观质量情况下,能够快速有效地实现EC。

Multi-view video plus depth （MVD） is a new 3D video format that would support 3D applica tion developed by MPEG. Such a format is a combination of texture video and associated depth maps. Consequently,for the efficient transmission of 3D video signals, the compression of texture and depth maps is required to reduce the inter-view correlation, the correlation between texture and the depth maps as well as the spatio-temporal correlation. Therefore, this makes the compressed bit-streams very sensi- tive to transmission errors and leads to error propagation easier. In order to prevent the error propaga- tion, effective error concealment （EC） algorithms are necessary. An error concealment method for multi- view video plus depth is proposed in this paper. Based on the corrections between depth map and texture map and the spatial corrections,the lost macroblocks （MBs） are classified into three types：simple MB, less complex MB and complex MB. Following the classification,corresponding operations,including can- didate motion vector refinement EC algorithm, depth-based outer boundary matching algorithm and error concealment with adjustable weights, are used to reconstruct the lost MBs. Experimental results show that the proposed method can effectively reduce the complexity of EC algorithm and ensure the video quality at the same time.