Research On Quality Inspection Of Wheel Casting Surface Based On Machine Vision

The data released by the relevant state departments show that the accident rate of motorcycles has increased year by year,and the traffic accidents caused by unqualified wheel hubs to the market account for a high proportion.At present,the main detection methods in domestic and foreign are radiation detection and manual detection.The radiation detecting device is cumbersome and harmful to the human body,and the accuracy of manual detection is low,which may cause secondary damage to the wheel hub.With the development of science and technology,the detection technology based on machine vision is becoming more and more mature.This detection method can ensure the consistency and accuracy of detection standards,and can achieve non-destructive testing.In the field of wheel hub quality inspection,a system that is simple and environmentally friendly and capable of detecting damage to the wheel hub is required.Therefore,using the machine vision method to realize the detection of the wheel hub defect is important to ensure the quality of the wheel hub and reduce the accident rate.Aiming at the defects of hub casting surface,the image system and defect detection system are designed by machine vision method with the existing experimental platform.The part to be detected is determined through the introduction and analysis of the morphological structure of the wheel hub.The causes of the common defects on the casting surface are analyzed,and the defects are classified into pits,lines and deviations.The imaging scheme is designed according to the composition of the entire detection system and the imaging module function,calculating the pixel size by image calibration,realizing the position and posture information of the best shooting point,and guiding the robot arm to image the wheel hub at the specified position.In the aspect of defect detection algorithm,the area to be detected is firstly roughly mapped by similar gray clustering method,the spokes are precisely positioned by vertical blur and edge detection,and the center circle is precisely positioned by extraction with roundness and circle fitting.A fuzzy adaptive threshold method with local characteristics is used to segment the pit defects,and then the defect regions are extracted according to the roughness and moment features.The line defects are extracted by the Steger method based on Hessian matrix,and then the anisometry features are used to extract the defect lines.A method of region shape matching is adopted for the deviation defect,the template area and the actual area to be detected are respectively positioned,and the defect area is extracted by making a difference between the two areas.Finally,all the extracted pits,lines,and deviation defect areas are determined whether it is a defect according to the manufacturer's defect size standard.On the platform,the imaging part can realize the omnidirectional imaging of the center circle and the spokes,the actual defect detection is carried out on the collected image samples,the correct rate of defect detection can reach more than 90%.It shows that the wheel hub defect detection system realizes automatic imaging,as well as the detection of defects such as pits,lines and deviations,and meets the design requirements.