A stereo vision measurement system Based on OpenCV

A complete realization method of stereo vision measurement system based on Open Source Computer Vision(OpenCV) was presented in this paper. Firstly, camera calibration based on 3D grid target was used to establish the relationship between the positions of the pixels and the scene points. This step was accomplished by OpenCV functions. Secondly, Harris corner detection was selected to get the feature points. Then, corresponding points between stereo images were calculated by sparse point matching algorithm. And epipolar constraint was added in the matching process in order to improve the accuracy and efficiency. Finally, 3D world coordinates of each feature pixel were derived from matching relationship and camera model. And then, the real size of the object was ascertained subsequently. Experimental results showed that the proposed method can build a complete and viable stereo vision measurement system.