The Research On Motion Compensation In DVC And MD

With the development of network and broad application of multimedia, many new applications arise, such as wireless PC cameras, mobile camera phones and multimedia sensor networks. They require easy encoding. However, the popular digital video coding standards, such as MPEG, H.263, H.264 which exploit complex motion estimation at the encoder, cannot meet this need. As new video coding paradigms, distributed video coding and multi-description coding are presented.In this thesis, the theory and the character of the distributed video coding will be introduced at first. Because of the importance of side-information to the exactness of reconstructing frames, the proposed motion compensated interpolation algorithm and the refined framework to side-information will be described in detail. Besides, using motion compensated interpolation to improve the side-channel in multi-description coding is also introduced.The main contributions of this thesis are as follows:1. A new motion compensated side-information interpolation algorithm is proposed. The characters of DVC are analyzed which are different from other video systems, and the motion compensated half-pixel interpolation is imported to improve the quality of the side-information in DVC.2. A new framework has been proposed with a refinement to side-information which is based on DWT. It has improved the estimation precision using multi-motion estimation selection with the help of the two frames which are transmitted and decoded by traditional channel.3. To improve the side channel quality, a new motion compensation interpolation algorithm has been proposed. Considering the motion in the sequence, the interpolation frame number is got and the interpolated frames are generated adaptively.