Stereo Matching Method In Stereo Video

Along with the development of science and technology, stereo vision technology has played an increasingly important role in modern life. Fields of application includes robot vision, object detection, virtual reality, scene roaming,3D reconstruction and aerospace. Research of this field has become a vital part of computer vision.In this thesis mainly discussed stereo matching technology. Other parts include related camera calibration and epipolar geometry theory, common type of noise and its processing method, the calculation process and category of disparity image, current research state of stereo matching algorithm and affect on efficiency and matching result with different size of matching window.A hybrid matching method in stereo video is proposed in this thesis. In the method area-based algorithm and feature-based algorithm are combined together, which leads to high increasing rate in efficiency and effect. Real time detection and satisfied results could be achieved using this method. In the conclusion part compares this method with other feature detection method.Much research on the improvement of algorithm efficiency and proposes a multi-thread model is done in this thesis. According to the intrinsic feature of area-based matching and thoughts of breaking up the whole into parts, this paper utilizes multi-thread to solve the problem. Matching efficiency is enhanced by50%.Stereo video matching algorithm and stereo matching efficiency enhancement method is realized using VS2010and OpenCV library. The method is modularized by dynamic link library technology aiming at improving the flexibility. An experiment platform for disparity video extraction is developed, which can be used to extract disparity video using stereo video and to enhance stereo video processing speed.