3-D Dimension Measurement Of Workpiece Based On Binocular Vision

In this paper, the three-dimensional measurement of workpiece is studied, and the left and right images of the target workpiece are captured by binocular camera. Firstly, the calibration principle and theoretical model of binocular camera are studied in detail, and the calibration of inside and outside parameters of left and right cameras is completed under the environment of Matlab. Secondly, Hough transform is used to detect the contour of the target workpiece after image preprocessing such as filtering, graying and binarization, and to extract its two-dimensional feature point information. Then, on the basis of epipolar rectification, there is only lateral parallax between left and right images, and Block Matching (BM) algorithm is used for stereo matching of relative images. After that, the depth information of the pixels is extracted from the disparity map obtained by stereo matching, and the dimension of the workpiece is measured by combining the two-dimensional feature point information extracted by Hough transform and the three-dimensional Euclidean distance calculation formula. Finally, the experimental results show the validity of the proposed three-dimensional measurement results based on binocular vision.