Real-time Detection System For Components Based On Machine Vision

With the rapid development of the electronics industry,the density of components integrated on printed circuit boards(PCB)is increasing.Once the components are found to be missing,mis-plugged,plugged in,etc.,the entire PCB board will be scrapped,seriously affecting the PCB.Production efficiency.Currently,the automatic optical inspection equipment for printed circuit boards(PCB)on the market is based on the quality inspection of solder joints,and the accuracy of detection of components on the surface of the PCB is low and easy to be falsely reported.Based on this,the real-time detection system based on machine vision in this paper can perform real-time and accurate detection on the circuit board components on the production line.The system includes collection and splicing section,PCB component detection section and data recording and display section.The main work of this paper is as follows:(1)Construction of hardware platform: box-type shadow-less light box,industrial computer with high processing capacity,industrial camera with high frame rate and high resolution,industrial lens with large aperture and mobile platform are used to realize the best environment for system operation.(2)Collecting and splicing board section: The section program is realized by the acquisition thread,the splicing thread and the sub-board thread independently.The program first determines the opening timing of the three threads by cyclically calculating the structural similarity change of the acquisition area in the PCB video picture.After the thread is started,the collection thread collects a small image of the acquisition area in the PCB video screen at a certain frame rate and stores it in a designated folder;after collecting a certain number of small images,the splicing thread uses the splicing algorithm to perform a fast real-time image.Splicing;When the PCB board enters the system seamlessly,the sub-board thread performs the sub-board through the sub-board algorithm,and the image after the sub-board isgradually sent to the detection queue for detection.(3)PCB component detection section: The section first corrects the PCB image to achieve coordinate mapping,and then performs component detection.Component detection is divided into presence detection and polarity detection.The component existence detection algorithm is designed with three detection algorithms: RGB image segmentation,histogram similarity and template matching.The component polarity detection algorithm is mainly for the detection of three polar components of diode,socket and electrolytic capacitor.The component polarity detection algorithm based on template matching and positioning,the electrolytic capacitor polarity detection method based on Hough circle positioning and the socket polarity detection algorithm based on threshold segmentation and positioning are designed.(4)Data recording and display section: The section first records the detected result data in real time,and then displays the detected result data to the user in the form of a Web page and a WeChat public number through the built Node server.This section allows users to view the result data detected by this system in real time.