Stereo Video Coding And Display Relevant Technology Research

In the next generation of video technologies, the stereo video as a digital medium which contains more information and data than ordinary video , has attracted more and more attention because of its intense characteristics of depth and third dimension. Stereoscopic video includes the depth information of scenes, which is more realistic in the characterization of natural scenes, and has unfolded the broad application prospect in the 3D television, 3D video conference with sound-surround ambiance and virtual reality. However, three-dimensional video has twice as much data as the monocular video, which creates difficulties for storage and transmission. Therefore, highly effective compression must be carried on. The paper first conducts a thorough research based on the encoding techniques of the H.264 stereo video and its associated reference model JMVM, and then analyzes the advantages and disadvantages of three traditional encoding schemes. On the basis of the result analysis, it concludes different application environment of the parallax estimation, the movement estimation and the union estimation. It also makes the main conclusion that the motion compensation mainly occurs in the motionless background area and slow-motion foreground area and occluded areas between two adjacent frames; parallax compensation is applied mainly in occlusion, deformation areas; joint compensation prediction mainly occurs in the edge of objects.The paper first conducts a thorough research based on the encoding techniques of the H.264 stereo video and its associated reference model JMVM, and then analyzes the advantages and disadvantages of three traditional encoding schemes. On the basis of the result analysis, it concludes different application environment of the parallax estimation, the movement estimation and the union estimation. It also makes the main conclusion that the motion compensation mainly occurs in the motionless background area and slow-motion foreground area and occluded areas between two adjacent frames; parallax compensation is applied mainly in occlusion, deformation areas; joint compensation prediction mainly occurs in the edge of objects.In accordance with human visual characteristics, the thesis analyzes the application possibility of the asymmetrical video encoding, and points out the questions which arise when the right perspective image refers to left perspective image in the encoding algorithm of the current asymmetrical three-dimensional video, and then proposes a fast matching algorithm - a direct reference frame prediction mode, which reduces the complexity of the code when the dynamic sampling of left perspective frame is carried on. compared with the existing algorithm. Simultaneously, the thesis analyzes distortion factors of the right perspective sampling image in the coding process of asymmetrical three-dimensional video, and proposes a more effective compression encoding algorithm of asymmetrical three-dimensional video, which can dynamically choose sampling scales according to the object code rate and find an optimum point between the degree of distortion and the code rate.The thesis also studies the real-time three-dimensional video display technology on the basis of the existing stereo video collection system, designs and implements a real-time stereo video collecting and displaying system based on the binocular video camera bumblebee2, which by using the latest DirectX 3D rendering technology, displays stereo video data collected by a camera on the ordinary CRT monitor by the color separation or in the time-sharing mode, so that users can appreciate the depth of stereo video to meet different demands, and also can save three-dimensional the acquisition as AVI-format video streams for further study. Finally, the thesis summarizes the work of the subject and shows the direction of further efforts.