Research On Key Technologies For Binocular Stereo Vision

Computer vision involves artificial intelligence, computer science, image processing, pattern recognition and many other interdisciplinary fields. Binocular stereo vision directly simulates the physical structure of human eyes, has choiceness characteristics such as small in size, low cost, little power consumption and high efficiency. Binocular stereo vision has broad application in the field of three– dimensional measurements. The research on binocular stereo vision has become one of the world's most popular research topics.This paper is focus on work piece's 3D reconstruction based on binocular stereo vision, it includes the research on image acquisition, camera calibration, stereo matching , 3D reconstruction and Portable binocular stereo 3D reconstruction hardware platform.This paper's main works are as follows:1.On the basis of analyzing the existing calibration method of binocular vision system and opencv camera distortion model, a new calibration algorithm is proposed. Very good results are obtained with real data calibration. The results meets the requirements of 3D reconstruction.2.After researching on the principle of SIFT stereo matching algorithm, we explicitly employ the color formation model with the parameters of lighting geometry, illuminant color and camera gamma in SIFT descriptor. So that SIFT descriptor possesses higher stability. In addition, for the building images with many areas that are locally similar to each other and areas with similar shapes but with different colors, the color and global information is introduced to improve the performance of SIFT descriptor. The improved SIFT algorithm can reduce mismatch probability of building images and improve matching results greatly. 3.Research and design of a mobile three-dimensional reconstruction of a portable binocular stereo vision hardware platform greatly improved the practicality of the system.4.Finally, using the experimental system of binocular stereo vision, conduct an experiment which includes following steps: images capturing, camera calibration, stereo matching and 3D reconstruction. Experimental results demonstrated the feasibility of related work by binocular stereo vision. It achieve the expected effect.