多视点视频加深度的容错编码与差错隐藏技术

项目名称： 多视点视频加深度的容错编码与差错隐藏技术

项目编号： No.61301112

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 无线电电子学、电信技术

项目作者： 马然

作者单位： 上海大学

项目金额： 24万元

中文摘要： 为了实现新型的3D视频应用，如3DTV和自由视点视频，在新一代的3DV数据格式中必须包括深度信息已成为共识。MPEG新近采纳了由纹理和深度图构成的多视点视频加深度（MVD）格式。差错控制是3D视频处理和传输系统的重要环节之一，对MVD的差错控制技术进行深入研究是十分必要的。 本申请项目主要研究基于编解码端的MVD的容错编码与差错隐藏技术。在编码端，充分研究MVD的时间、视点和深度的不同重要性和相关性，拟建立可分级判定模型、相邻视点运动矢量相关性模型和视频图/深度图相关性模型等，从而实现多重分层编码，以提高容错性能。在解码端，拟基于多重的空间、时间、视点和深度相关性，确定候选矢量集，以准确迅速地实现丢失宏块的恢复；并在此基础上加入静止宏块预判定、边界宏块预分割等以提高隐藏的速度和效果。最终，实现完整的容错编码与差错隐藏的联合方案以实现有效的网络传输。

中文关键词： 多视点视频加深度；容错编码；差错控制；可分级编码；差错隐藏

英文摘要： In order to enable new 3D video applications, such as 3DTV and Free Viewpoint Video, it is a common idea to include depth information in the upcoming generation of 3DV data format. Multi-view video plus depth (MVD) format, which consists of texture and depth images, has been recently specified by MPEG. For error control is a key step in the whole system of 3D video processing and transmission, error control for MVD must be studied deeply. This project will mainly study on error resilient coding and error concealment for MVD based Codec.At the encoder, the model of relevancy between video and depth images and the model of relevancy among adjacent views will be built based on different importance and relevancy of time, view, and depth for the MVD format.So, multiple layered coding can be implemented to conform to the various bit rates.At the decoder, the set of candidate vectors can be decided based on the relation of space, time, view and depth, such that the lost Macroblocks are recovered accurately. Furthermore, the pre-decision of still MB and pre-segmentation of boundary MB are combined to improve the efficiency of error hiding. Finally, the complete sheme of resilient coding and concealment based on the codec can satisfy the request of network transmission.

英文关键词： Multi-view video plus depth；Error resilient coding；Error control；Scalable coding；Error concealment