Research For Shape Detection Technology Based On Binocular Vision

With the rapid development of modern detecting technology, non-contacted method has been widely used in many areas such as industrial inspection for its high detecting speed and no requirement of objects’material. In nowadays, machine vision technology is the most popular one in non-contact detecting technology.This paper mainly discuss the stereo vision technology. By using two cameras capturing the image, the disparity can be generated just like human eyes. Then the stereo matching algorithm is used to match the two images and the disparity map can be generated. On the bases of the disparity information and triangle measuring principle, the depth information of the object can be extracted. Parallel binocular vision system is adopted in this paper to capture images and the regionbased stereo matching algorithm is mainly discussed. At the meantime, how the match window can affect the result of stereo match is proposed and the correlated experiments have been done to confirm the principle. Based on the experiment result, an appropriate window size is chosen to use in this paper. Because the key to the shape detection is depth extraction accuracy, which is definitely dependent on the correlation match result, three validation methods have been proposed to decrease the probability of mismatch and as a result optimize the stereo match algorithm. At last, the3D position of the points on the object surface can be got, based on which the three dimensional reconstruction has been done and the3D graphic is exported.