Research On 3D Measurement Method Of Polyhedron Connectors Based On Binocular Vision

With the continuous development of manufacturing industry,the traditional contact measurement begins to restrict the production efficiency.And it is particularly important to improve the speed and accuracy of measurement.This topic is to carry out research under this background,and study the method of on-line rapid measurement of connectors by using binocular vision measurement technology.This paper is based on the 3D reconstruction of binocular stereo vision to measure the 3D size of the polyhedron connector model.The research is mainly carried out from four steps: binocular camera calibration,image preprocessing,stereo matching algorithm and 3D reconstruction.The main research contents and work are as follows:(1)This paper studies binocular stereo vision,expounds its imaging theory,and deduces the formula involved.Several common camera calibration methods are compared and zhang’s calibration method is selected to calibrate the internal and external parameters of the camera.The calibration of binocular camera is completed by programming with C ++ and Open CV,and the improved genetic algorithm is used to optimize BP neural network to further optimize the calibration results,so as to improve the calibration accuracy.(2)Taking into account the image distortion caused by external factors during the production process of the camera,it is necessary to perform stereo correction on the image to remove the distortion.Various noises may appear in the imaging process,and the noise needs to be eliminated.Mean filter is used to remove the sharp noise,and the median filter is used to remove the salt and pepper noise.Then,the Canny algorithm is used to extract the edge information.(3)In the aspect of stereo matching,a variety of stereo matching algorithms are summarized,from the comparison of matching effect and matching time,a semi-global stereo matching algorithm is selected for research.Moreover,the constraint criterion and four steps of semi-global stereo matching are described,which are matching cost calculation,cost aggregation,parallax calculation and parallax optimization.Aiming at the problem that Census transform relies too much on the center pixel,this paper studies the method of Gaussian weighting to connect all pixel values in the matching window and combine the original color information of the image,so as to make the new center value more reliable.(4)In the aspect of 3D reconstruction,the disparity map generated by stereo matching is combined with the calibration information to perform three-dimensional reconstruction of the object to obtain the point cloud image of the object.According to the point cloud image,the 3D coordinates of the object are obtained and the 3D sizes are calculated.The reliability and stability of the algorithm are verified by experiments.