Research And Application Of Hole Position Measurement Of Riveted Parts Based On Binocular Vision

At present,the factory's existing small semi-automatic riveting machine feeding device,which needs to input the hole position information of the sheet to be riveted manually.This kind of positioning method has low efficiency,poor universality,and can't meet the standard of full automation.In order to achieve the integration of vertical rivet machine and small semi-automatic feeding device,reduce the labor intensity of riveting process staff and improve the riveting efficiency and quality,this topic focuses on the algorithm of measuring the riveting hole position information using binocular stereo vision technology.The main research contents of this subject are as follows:(1)Calibrated binocular camera.Analyzing the transformation relationship between each coordinate system and the camera imaging model.Binocular cameras are calibrated based on the Zhang Zhengyou calibration method and the binocular vision principle.Through the analysis and verification of several groups of calibration experiments,it achieves that the calibration error is 0.22 pixels and gets the internal and external parameters of binocular camera.(2)Extract and match the feature points of riveting sheet image.The maximum value method and the median filter are used to complete the pre-processing of the acquired image.The neighborhood detection method is used to connect the contours detected by the Canny operator.Combined with the characteristics of the sheet,a linear constrained image segmentation method based on hough transform is proposed to completely segment the riveting sheet and the positioning rectangular block.A matching method based on position constraint is proposed to match the extracted feature points one by one.(3)Unify coordinate system.The riveting hole is transformed into a global coordinate system established on the positioning right angle block by the pose description,where the relative pose parameter R,T is solved by SVD(Singular Value Decomposition)to achieve the purpose of pose estimation.(4)Build the system experiment platform.Select the appropriate hardware for the rivet hole binocular vision system and build a binocular vision platform,design a reasonable system interface,and combine the proposed algorithm to achieve the location of the riveting hole.The experimental results show that the binocular vision positioning technology can accurately locate riveting holes.The method is simple and inexpensive.The absolute average error of riveting hole center positioning based on binocular vision is 0.1584 mm,and the absolute average error of vertical z-axis depth information is 0.0403 mm,while the positioning accuracy of riveting center of thin sheet in riveting industry is within 0.2 mm.This method meets the industry requirements.