Research On Pins Detection Method Of Automobile Connectors Based On Binocular Stereo Vision

By the end of 2021,the number of automobile registration in our country has reached302 million.The automobile connector is one of the core parts of the automobile,which is responsible for the flow of electrical signals in the various parts of the automobile,and during mass production processes,the heights of pins may be too long or too short,leading to the abnormal communication of the automobile connect,which directly affects the safety and stability of the automobile.Aiming at the disadvantages of low efficiency of traditional pins detection and easy damage of contact detection,a binocular stereo vision based detection method of automobile connector pins is proposed.Through the binocular camera calibration,image preprocessing,target recognition,feature point extraction,stereo matching and other aspects of the research,the binocular vision technology is used to reconstruct the depth information of the automobile connector pin,and the quality discrimination.At the same time,corresponding improvement plans are proposed to address the problems of interference from edge shadows in the actual work environment of automobile connectors,small illuminated areas of the pins,and disorder of stereo matching primitives.The specific work content and innovations are as follows:(1)Research camera imaging models and camera calibration.Firstly,the mathematical modeling of camera imaging process is carried out to study the transformation relationship between each imaging coordinate system.At the same time,the nonlinear imaging model is introduced to quantitatively analyze the image distortion caused by lens distortion.The binocular stereo vision model is further studied,and a set of binocular stereo vision hardware platform is built.The binocular camera was calibrated by Zhang Zhengyou calibration method,and the internal and external parameters of binocular vision system were obtained.(2)Research on automotive connector Recognition method based on geometric features.An automobile connector recognition method based on improved Hu invariant moments was proposed.Based on the traditional Hu invariant moments,by constructing a gain function to improve the description function of Hu invariant moments on contour features,and an adaptive pyramid search strategy was proposed to improve the speed of template matching.The automobile connector is identified by template matching technology as an area of interest to remove the shadow interference caused by the reflection of the plastic shell in the edge area of the automobile connector.(3)Research on stereo matching method based on feature points of the pins.In this paper,a feature point extraction method based on centroid fitting is proposed.The key feature points of pins are extracted for stereo matching.Introducing morphological operators to enhance pin features and solve the problem of mis-extraction caused by small light receiving areas of pins.Further extracting pin features through binarization processing,combined with connected domain area analysis to screen out accurate pin contours.And a sort method is designed to automatically sort the pins and the three-dimensional coordinates of the pins are reconstructed by matching the centroids of the pins.(4)A software platform for the detection of automobile connector pins was constructed.Based on Windows10 operating system,Qt Creator is used to invoke Visual Studio 2019 compilation environment,and Open CV4.3.0 visual library is combined to realize software platform development.The accuracy and stability of the algorithm are verified by the convenience of the software platform.The experimental results show that the overall detection time from reading binocular images to determining whether the pins are qualified is within 1 second,and the dynamic repeatability error of pin reconstruction is 0.354 mm.