Research On Some Problems Of Stereo Video Coding

Stereoscopic video technology is a new video technology which can provide stereo feeling. It is an important development direction of future multimedia technology. The shortcomings that large amount of data, inconvenient storage and transmission extremely limit its wide range of applications. Therefore, how to efficiently encode stereo video will be the focus of future research. In this paper, we did some research on estimated parallax in three-dimensional video, stereo video object segmentation and3D video coding framework.The disparity estimation is an effective method to remove spatial redundancy of three-dimensional video data. It can be effectively to encode and compress stereoscopic video data. So. it is one of the key techniques in stereo video coding. This paper systematically introduces a series of related algorithms and research results of disparity estimation. Especially, disparity estimation based on dynamic programming algorithm and based on belief propagation disparity estimation algorithm, and its improvement program for a number of related experiments. It shows that these algorithms have their own advantages and disadvantages.Object-based coding to reduce coding errors, improve the coding efficiency and visual effects, to the scene for structural description also has excellent performance, adapt to the new application fields. Video object segmentation is the basic of the object-based stereoscopic video coding. Therefore, the quality of video object segmentation affects the pros and cons of the three-dimensional object-based video coding performance, and it is one of the key technologies of the stereoscopic video coding to segment to video object segmentation. In this paper, we put forward a method that combines edge with depth information for stereoscopic video segmentation after the analysis of existing segmentation algorithms. Through the Canny operator edge detection, matching combined with fixed window iteration to get the ultimate segmentation result. The experimental results show that the method works well. This is also an innovation of this paper.Also, the paper introduces algorithms and frameworks of the stereoscopic video coding systematically, and focuses on the H.264-based of stereoscopic video coding three scheme, then some related experimental simulation did. The experimental results show that, the right channel using only of the DCP encoding the worst effect of the scheme. At low bit rate, the right channel MCP+DCP coding the program significantly better than the left and right channels independently coded scheme. The greater the bit rate, the right channel MCP+DCP encoding program advantage is the less obvious. Finally, a brief summary of scalable coding scheme and video quality assessment.