| The3D (three dimensional,3D) video technology can provide viewers an immersive feelingwith the depth information in the natural scene. And it has been applied to3D cinema, stereo videoconferencing, remote expert consultation, as well as navigation and mobile video services.Recently, Europe’s Seventh Framework Programmer for Research and Technology started theobjective of Mobile3DTV in the context of the Information&Communication Technology (ICT),which lead to the stereo video technology based on mobile terminal to a hot spot of research andapplication. The mobile channel has multipath propagation, strong ability of noise interference, andthe complex environment of terminal usage, which makes mobile bandwidth variation relevantlyfierce. Compared with other transmission channels, the mobile channel transmission may lossmore packs, resulting in a mosaic decoded video, black block, green block, pause and so on. As forstereo video sequences, due to reasons of inter-view reference, the error of one channel will affectthe video qualityof other channels. What’s worse, the error diffusion will cause a serious decline inthe quality of whole3D video. Therefore, error concealment (EC) technology oriented the Mobile3DTV stereo video becomes more and more important.This dissertation firstly gives a briefly introduction of H.264/AVC video coding standard andthe Mobile3DTV standard. Then it summarizes and analyzes the Mobile3DTV stereo videocoding scheme. And at last, it encodes the different presentation formats, which defined by Mobile3DTV through the different encoding structures.(1) In order to low the computational complexity of the Conventional Stereo Video errorconcealment algorithm, the Stereo SEI structure of the Mobile3DTV standard is used. In thedecoder, according to different situations, different error concealment schemes are used toconcealment loss pixels. The results of this research show that the proposed scheme can effectivelylow the computational complexity of the error concealment processing. The proposed algorithmcan gain the value of PSNR from0.16dB to3.4dB, compared with the frame copy algorithm.(2) As for the Mixed Resolution Stereo, this paper establishes a multi-resolution encodingplatform based on the3D video checksum platform. Left view images use the full resolution, whilethe right view images firstly use the down-sampling resolution, then co-encode with the left view, concealing all the lost macro-block in the decoding side, and at last up-sampling the right viewCompared with SAD algorithm, the proposed algorithm effectively reduces the matching numberof macro-blocks and lows the computational complexity. Proposed algorithm can gain the value ofPSNR from0.20dB-8.33dB, compared to the frame copy algorithm.(3) For the problem of some macro-blocks loss or consecutive macro-blocks in Video plusDepth Representation of Mobile3DTV, different error concealment schemes are used in thedecoder for the color and depth videos. Experimental results show that, compared with the framecopy algorithm, the proposed algorithm can gain the value of PSNR from0.17dB to5.48dB, andfrom-0.19dB to1.93dB compared with boundary matching algorithm. And the depth image theproposed algorithm the same to copy algorithm. The computational complexity of the proposedalgorithm is between direct copy and direct boundary matching algorithm.All the error concealment schemes take the computational complexity into consideration. Thevideo recovery PSNR value is maximized as possible under the premise of lowing thecomputational complexity, for the purpose of ahcieving a better adaptation to the application of themobile terminal. |
PDF Full Text Request