Research And Design On PCB Defects Intelligent Visual Inspection System

In recent years, with the improvement of the modern information level, peoplehave more and more requirements for the quality of the electronic products. As thecarrier of electronic products, the quality inspection of printed circuit board (PCB)has also become a core issue in the electronics manufacturing industry. Thetraditional PCB defects detection is achieved by artificial detection or functionaltesting, but these methods have been unable to meet the testing requirements forhigh-speed and high precision in the production line. In order to improve the speedand accuracy of defects detection, the paper designs a PCB intelligent detectionsystem based on machine vision, and realizes the online real-time detection for PCBdefects.Firstly, the article analysis the significance of researching about PCB defectsintelligent inspection based on machine visual by introducing the present situation ofthe development of the PCB defects inspection at home and abroad. Then, it designs aset of PCB defects intelligent visual inspection system and gives its general structureand operating principle. Through the research of the system in three aspects includingOptical imaging technology and Machine system and Electrical control system, thekey technology of the PCB defects intelligent visual inspection system has beenunderstanded in a more depth.Secondly, aiming at the factors that affecting the quality of the collected images,the image pretreatment algorithm is analyzed. After the gray image contrastenhancement, the paper adopt a kind of image denoising method to remove thescattered noise and gaussian noise by combined with the median filter and bayeswavelet denoising filter. Based on the full consideration of the relationship betweenthe grayscale and space and the algorithm’s complexity, the paper puts forward arapid iteration algorithm of two dimension Otsu threshold segmentation, which couldsegment the background area and thevtarget area precisely.Then, it detects and recognizes the PCB defects. The article mainly puts forwardtwo methods to detect and recognize two types of different PCB defects: one methodis using detecting PCB XOR standard PCB or false standard PCB algorithm to detectthe defects such as short circuit, open circuit, burr and so on; Another method is fuzzyneural network classification algorithm which is used to detect the solder is normal or not.Finally, the article designs and develops control software of the system, which isproviding intelligent human-machine interface, solving the problems in the practicaltest and ensuring that the system can run fast and accurately, then accomplishing thePCB defects online real-time detection.