| The vision is the most important source of human access to information while the video holds an important position of the multimedia. However, there are many problems in the video’s storage and transmission, for example, low network bandwidth, taking up too large space, network error loss. Without changing the existing hardware condition, the reduction of the redundancy and using the fault tolerance are the key to solve the problem.H.264/AVC is the latest generation of video coding standard jointly developed by ITU-T and ISO/IEC. Compared with the previous standard, it can save50%of bit rate with similar visual quality, due to using smaller block mode, multi-frame reference, FMO and other technologies. On the other side, because of the signal instability and other reasons, there are errors or packet loss in the transmission process of the compressed video stream. Error concealment (EC) can repair damaged video sequence based on the existing information.Frames in H.264/AVC can be divided into intra-coded frames I, SI and inter-coded frames P, SP B, and SB. In this thesis, for P-frame we found that the performances of original boundary matching algorithm (BMA) and the Lagrange interpolation (LI) error concealment effect are not satisfactory. So we propose a new method. Additionally, the proposed method can be easily extended to other inter-coded frames.The emergence of3D video brings the error concealment technique great challenges. On the other side, the3D video provides more useful information.3D video can be divided into several types, Multi View Video (MVV), Conventional Stereo Video (CSV) and Depth-Image Based Rendering (DIBR). This thesis focuses on error concealment for DIBR video, in which we have proposed a new error concealment method for3D video based on the depth information.In conclusion, to solve the problems above, new algorithms are proposed for both2D and3D video, which can significantly improve the performance of error concealment. |
PDF Full Text Request