Research Of Multi-view Auto Stereo Image Synthesizing Algorithm

Multi-view auto stereo image synthesizing algorithm, which directly influences the development of taking, transporting and displaying technique of auto stereo image, is a key technique to achieve the industrialization of multi-view auto stereo image. Although multi-view auto stereo image is being applied in some special field, but many other application fields have been flinched for its high cost. If multi-view auto stereo image synthesizing algorithm can be applied in practice more or be introduced into related research domain of taking and transporting of auto stereo image, it will effect auto stereo image development deeply. The research of Multi-view synthesizing algorithm is still on the step of developing, which cannot solve the problem of quality and efficiency of the new novel. If high quality is needed, the algorithm is rather complex; but if real-time disposal is needed, it must use easier efficient synthesizing algorithm to improve image synthesizing efficiency.According to the requirement of auto stereo image development, this paper firstly introduces the development state of stereo image, which begins at stereo visual theory and then gradually gets into taking and displaying method of stereo image. Secondly analyzing the geometry principle of stereo image, the paper presents the geometry model of stereo image via mathematic modeling, which is the basic of algorithm being discussed below. Afterward, Warp algorithm and Remedy algorithm that are both high quality multi-view synthesizing algorithms are discussed here, and also the algorithm principle is dissertated. But two algorithms are both complex and the Remedy algorithm can be only used into dealing with black and white image. Therefore the paper carries out AIV algorithm, Grad algorithm and Self -adaptive Shift Parallax Loading algorithm, which are all efficient algorithms. Three algorithms are in character as simple disposal process, specially the Self-adapted Shift algorithm is a method of 2D to 3D with multi-view. They all have wide development space. Lastly, the paper discusses the experiment method and process under Visual Studio6.0 and Matlab environment, subjective test and objective test. The result of tests is satisfied.