Research On Detection System Of Hub Bearing Flange Based On Machine Vision

Hub bearing flange is the key part of the third generation hub bearing unit with anti-lock brake system,and its thread hole and appearance inspection are strictly required.The traditional manual detection method has low accuracy,slow speed and high cost,which can not satisfy the current detection requirements.With the continuous development of machine vision technology,it has been widely used in industrial detection system due to its high accuracy and high speed.This subject aims at the existing drawbacks and combined with industry standards,then put forward a flange detection system for hub bearing based on machine vision.The main content of this paper is as follows:(1)Two detection stations are designed by using photoelectric sensor to trigger camera to collect images.They are used to detect the deviation of threaded holes,the integrity of threaded holes and the rust spots on the surface of the workpiece.Two sorting channels are designed with push rod device,and the qualified and unqualified workpieces are separated according to the test results.(2)The selection of camera,lens and light source in the image acquisition module is expounded in detail,and the optimal imaging system is designed for image recognition.(3)According to different detection requirements,the algorithm ofeach detection part is designed innovatively.Aiming at the problem that the traditional correlation method takes a long time to do template matching,the quadratic template matching algorithm with error threshold is introduced,and an improved method based on contour template matching positioning is proposed.The speed of template matching by this method is about 10 times higher than the original,and the time required is shortened from 30 ms to 3ms.Compared with the original method,the speed of template matching is increased by 10 times,and the needed time is shortened from 30 ms to 3ms.Pointing at the problem of large error in least squares circle fitting with large interference sample points,an improved least squares circle fitting algorithm based on randomized pre-fitting is proposed.By comparing the results before and after the improvement,it can be seen that the improved fitting algorithm makes the detection results more accurate.(4)Using Visual Studio programming platform,C# combined Halcon mixed programming,designed the upper computer UI interface to facilitate system operation.(5)Experiments were carried out on each detection content,and the specific experimental results were obtained.The experimental results show that : The detection accuracy of workpiece thread integrity is88%,the detection accuracy of surface rust defect is 92%,and the detection accuracy of offset degree of threaded hole position is 96.4%.