Research On Surface Defect Detection Technology Of USB Plug Based On Machine Vision

With the rapid developments of science and technology,the automated development of production has been gradually matured in some area,automated assembly line can greatly increase the production efficiency and reduce the labor costs.However,the surface defects detections of USB plugs are still using the human eyes,because of the difficulties of the defects acquisition and detection.The surface defects detections using eyes are not suitable for the automated assembly line because they are not objective,their higher costs,lower efficiency and lower accuracy.In this paper,we propose a surface defect detection technology of USB plugs based on machine vision.The main work contents are as follows,1.The background,significance and the expected target of this paper are introduced.The traditional surface defect detection methods,such as magnetic particle detection,penetration detection,eddy current detection,ultrasonic detection are introduced.By comparing the relative merits and using range between different methods,we choose the detection method based on machine vision.2.The overall structure of the surface defect detection technology of USB plugs based on machine vision is presented.The principle of surface defect detection method based on machine vision is introduced.The parameters of the camera and the lens are determined and verified according to the system requirements.The advantages and disadvantages of the commonly used image processing software are researched.HALCON is chosen as the processing software in this paper.3.The illumination method and the light source of the machine vision system are designed.The commonly used illumination ways and light sources of the machine vision system are introduced.The improved coaxial light source illumination and the ring light source illumination are researched.The ring light source illumination is chosen because of its sensitivity and good contrast in surface defect detection of USB plugs.4.An improved hysteresis threshold algorithm is proposed.The algorithm can automatically choose the appropriate threshold for different defect images.This paper has verified the space and frequency characteristic of the scratch defect,and presented a scratch defect detection method based on Gabor filter.This method can effectively suppress complex background noise of the USB plug,and improve the sensitivity for the scratch defects.By bulk USB samples validation using this algorithm,the defect extraction accuracy reaches up to 95% and the failure rate is 1% for plane defect extraction,and defect extraction accuracy reaches up to 92% and the failure rate is 4% for scratches extraction,which reaches the expected goal.5.The Support Vector Machines(SVM)classification algorithm for the surface defect detection of USB plugs is presented.The defect classification model and commonly used classification algorithm are introduced.The SVM classification principle is introduced in detail,which can analyze the different geometric characteristics of holes,scratch,scratch defect using area,roundness and gray average features.By bulk USB samples validation using this algorithm,the classification accuracy of SVM method is 96%,which reaches the expected goal.