A Quality Inspection System Of Printed Circuit Board

With the rapid development of modern circuit board manufacturing industry, circuit board products tend to be more complex and precise, and traditional circuit-board testing system falls far short of the requirements for testing accuracy and reliability of the circuit board quality. So,intelligent quality detecting system of circuit board has become an urgent need.For this situation, according to the principles of optical inspection and functional testing,the paper introduces a circuit board quality inspection system with the use of machine vision technology. The system can automatically and stably implement the detection of defects of the circuit board in real-time, including detection of defects on the surface of the circuit board components and the display defects of LED screen. In this paper, design of hardware and software is confirmed and implementation of image detection algorithm according to the overall design architecture which is analysed by the function of the system.Experimental results show that the method proposed in this paper is feasible and effective.The main research contents of this system are as follows.(1)Hardware platform design mainly includes image acquisition module and LED driver module. Image acquisition module consists of high-pixel industry camera whose model is VT_EX500CPS, VT_LEM0616MP5 lens and COP-150 W coaxial light source. In order to ensure the quality of the image, the parameters of the camera should be set to an proper value,so the image of circuit board can be captured and transformed in real-time; LED driver module is mainly composed of PST series programmable power supply and RS232 serial circuit and power supply circuit, this module is used to ensure that when detecting LED screen, it can provide power to the circuit board and the LED screen will light up.(2)Software platform design, it mainly includes the image acquisition software, LED detection software and components detection software.According to the processing flow of each software, it completes the design of interface and the framework of software by the Labview programming and realizes the image processing algorithm by Matlab programming.(3)Image processing algorithms, in this part, image preprocessing, feature extraction, ROIextraction and defect detection algorithms, etc. are included. Image preprocessing, mainly includes image graying, smoothing and enhancement processing; feature extraction is mainly based on Harris algorithm; ROI extraction is based on reference point which is determined by right corner coordinate got by feature extraction and image features, which is used to extract the component of the circuit board.(4)About defect detection, this paper uses the image matching, image constriction, threshold segmentation, Hough transform algorithm, etc. to implement the detection of circuit board intelligently. In detecting the display defects of LED and the defect of component which is forgotten to be welded, grayscale correlation-based template matching method are used, and then use simplified SSDA matching algorithm to calculate the similarity, and finally determine whether the defect exists by using the experience threshold. Meanwhile, apply nonequivalence operation and morphology Corrosion processing to the image in the last step to obtain the final image with defects. In the detection of the capacitors which are welded at wrong position, a method which is based on adaptive threshold segmentation is present in the paper. Also, in the detection of the chips that are welded at wrong position, a method by using the Hough transform with the same step is proposed to judge the welding direction of the chips.(5)In the end of the paper, it analyzes the processing results of the system by using Circuit boards with different types of defects and uses a certain number of each kind of circuit board to obtain experimental results. According to the summary and analysis of the experimental results,it can be concluded that the detection time of the system is less than 1 second, the detection accuracy is above 90%, which proves the feasibility of the system.