Design Of PCB Inspection System Based On Machine Vision

In the production process of printed circuit board(PCB),the defect detection of components is an important part to ensure the qualified rate of PCB.Due to the longtime and high-intensity manual operation,the traditional manual detection method is prone to make mistakes,and the detection efficiency is low,which often can not meet the detection requirements of PCB production.However,the automatic optical detection equipment with high precision and high stability is very expensive,which only a few companies with strong economic strength can afford.Therefore,it is of great significance to develop a set of PCB detection system with low cost,high precision and good stability.This paper mainly solves the problems of the existence detection and polarity detection of the component on PCB.The image tracking algorithm,the component existence detection algorithm and the polarity detection algorithm are studied.A set of PCB board detection system based on machine vision is designed,which realizes the rapid and efficient detection of PCB on the production line.The work contents of this paper are as follows:(1)In PCB image tracking algorithm optimization: when tracking uneven illumination PCB,the stability of pyramid matching algorithm based on gray feature is very poor.Secondly,the real-time performance of tracking PCB is also poor,which can not meet the high real-time requirements of the scene.In order to improve the real-time performance of the tracking algorithm,the image is downsampled and only the filtered edge information is used as the feature optimization algorithm;In order to improve the stability of the tracking algorithm,a strategy optimization algorithm is adopted,which first determines the candidate matching region by fast rough matching,and then fine matching the candidate region,in order to improve the stability of the tracking algorithm.By tracking 852 PCBs of three models,the accuracy of the two algorithms is 98.12% and 99.53% respectively.Compared with the former,the accuracy of the optimized algorithm is improved by 1.41%,and the time consumption is reduced by more than half.The performance and real-time performance of the algorithm are greatly improved.In addition,according to the different pixel size of PCB,two tracking modes are designed,the first is single board tracking,the second is double board tracking.(2)The image coordinate calibration algorithm is studied:Due to the transmission speed of conveyor belt,external noise interference and image matching error during tracking,the component position coordinates on the standard template can not be accurately mapped to the PCB test diagram.Therefore,this paper proposes a Hough matching image coordinate calibration algorithm based on edge curve.The algorithm takes the extracted edge curve as the feature and combines with Hough matching search strategy to accurately find the coordinates of the matching area with the logo image on the PCB test diagram.Based on the coordinates of the matching area,the PCB test diagram is calibrated.The experiment shows that the algorithm can effectively calibrate the PCB test diagram.(3)In order to solve the problem of inconsistent gray features of tiltable components,:the algorithm of image chromatographic similarity is used to detect the existence of components through image chromatographic similarity.If the color of components is similar to the background color of PCB board,the background color of PCB board will produce interference when detecting the existence of components.In order to avoid interference,an algorithm based on gray feature matching is adopted to distinguish the background of PCB board and components by using gray feature,so as to eliminate the interference of background.In order to test the effectiveness of the detection algorithm,6784 components are detected.The experimental results show that the detection algorithm can accurately detect the existence of components,and the detection accuracy is as high as 99.21%.(4)The polarity detection algorithm of components is studied:In order to solve the interference of uneven illumination and silver white characters on diode polarity detection,an algorithm based on Niblack local threshold segmentation is proposed.The gray transformation and background correction methods are used to reduce the influence of uneven illumination;Niblack local threshold segmentation algorithm is used to eliminate the interference of silver white characters on diode polarity detection.In order to solve the problem that PCB background and metal sequins are too dark to interfere with the polarity positioning of electrolytic capacitor,a method based on edge curve detection and circle fitting is proposed to limit the detection area of edge curve,so as to eliminate the interference of PCB background;The method of edge curve detection and circle fitting is used to eliminate the interference of too dark metal sequins.Aiming at PCB background interference socket polarity location,an algorithm based on threshold segmentation and edge detection is proposed.Threshold segmentation is used to separate socket and PCB background and eliminate background interference.In order to test the effectiveness of the proposed polarity detection algorithm,the polarity of 3486 diodes,2488 electrolytic capacitors and 1780 sockets are detected.The experimental results show that the detection accuracy is 98.94%,98.36% and 99.13%respectively.Compared with the template matching algorithm and Otsu threshold segmentation algorithm,the proposed algorithm improves the detection accuracy by2.82% and 4% respectively,Compared with ght algorithm and random Hough transform algorithm,the accuracy of the proposed algorithm is improved by 3.8% and1.54%.(5)In the aspect of performance verification of PCB detection system: the detection accuracy of the detection system was tested,and six batches of different PCB boards were tested in real time on the production line.In this test,92946 components were detected,4 false positives,and the total false positives rate was 0.0043%,0 false positives,and the total false positives rate was 0,PCB detection system meets the detection requirements of actual production.