Researches On Several Key Techniques Of Three Dimension Video Coding

Recently, with the development of Computer Graphics and Computer Version, techniques toward three dimension (3D) video have become a hot topics in the video coding area. Meanwhile, the great success of stereo video such as Avatar indicates the huge market potential of the 3D video. Since the huge mount of 3D video, it is important to develop more efficient compressing techniques for 3D video. In actual situation, the condition of network bandwidth, capacity of client is different; Therefore, researches on the scalability of 3D video coding are valuable in both of the theory and application. As an important branch of multiview video coding, asymmetric stereoscopic video coding attracts more and more attentions because of its higher compression efficiency and approximate subjective quality compared to traditional stereoscopic video. Based on these properties of asymmetric stereoscopic video coding, researches on the scalability of stereoscopic video coding are of both theoretical significance and prctical value. As a important part of 3D video, depth video occupies many proportion in 3D video bitstream. However, the existing coding techniques are designed toward texture video and do not fully consider the characteristics of depth video. Therefore, researches of depth video coding are of both theoretical significance and practical value.Based on the properties of asymmetric stereoscopic video coding, the thesis investigates the scalability of asymmetric stereoscopic video coding by combining techniques of the inter-view prediction and inter-layer prediction. In addition, the thesis analyses the characteristics of depth video and investigates the techniques for depth coding. The main contents and novelties of the thesis are as follows:1. The thesis proposes a new coding structure of asymmetric spatial scalable stereoscopic video coding.Combined the advantages of asymmetric stereoscopic video, such as high compression efficiency and approximate subjective quality compared to traditional stereoscopic video, the thesis proposes a new coding structure of asymmetric spatial scalable stereoscopic video coding with the help of the scalable coding theory of the scalable video coding. The thesis introduces spatial scalability in the non-base view coding and enables inter-view prediction in all of the spatial layers. The coded bitstream can be decoded as asymmetric stereoscopic video or symmetric stereoscopic video based on different conditions and preference. The proposed method improves coding efficiency of stereoscopic video as much as possible while providing resolution scalability.2. The thesis investigates and improves the techniques of depth video coding.Based on the characteristics of depth video and disadvantages of existing coding techniques, the thesis proposes improved methods for depth coding aiming at these problem. The thesis investigates the mode decision process of depth coding and discusses spatial adaptive quantization for depth video coding.Experiment results show that, the proposed asymmetric spatial scalable stereoscopic video coding scheme has significant improvement in the coding efficiency compared to other coding schemes providing similar properties; Moreover, the depth coding schemes proposed in the thesis improve the coding efficiency of depth coding.