Research On Defects Detection Technology And System Of Printed Circuit Board

In the modern electronic industry, Printed Circuit Board(PCB), which is the supplier of electronic components' electrical connections, has been widely used in various fields, meanwhile the quality of the on-board component' placement will directly affect the product's performance. In recent years, with the high complexity of the PCB and high increasement of the yield, due to the shortcomings of intensity of labor, poor reliability, low efficiency and high cost, artificial visual detection and other traditional PCB defect detection methods can not meet the needs of modern PCB defect detection. Therefore, studying and developing an automatic defect detection system, which can replace artificial visual detection, has great realistic significance and practical value.According to the theory of visual detection, using image processing technology, this paper designs a software program for PCB defect detection. PCB defect inspection system mainly has three parts, motion control device, image acquisition equipment and image processing software system. And the image processing software is the study core of this thesis, the paper mainly studies algorithms of the detection software and applications of the key functional modules, its detection performance of each functional module is verified by VC++program. The software detection section includes several module, such as image pre-processing, threshold segmentation, image comparison and defect recognition.In each software module, this thesis mainly studies the design and realization of algorithms in the process of PCB defect detection. In the module of image pre-processing, combined with PCB's feature of having dim color, the PCB picture may be mixed with noises in the image acquisition process, so this paper uses image enhancement, image denoise and pre-processing means to gain high-quality PCB image. In the module of threshold segmentation, because of PCB's gray histogram having the characteristic of twin peaks, this paper chooses the petronas twin method of one threshold segmentation method to gain PCB binary images of clear characteristics and low noise in the current maturity theory of threshold segmentation algorithm. In the module of image comparison, locate the two images in space, and make them have an XOR operation to gain the defect image, in order to exclude the existence of pseudo-defects, the paper uses morphological methods for further processing, and then obtains the defect characteristics of the tested PCB image, in the end we get the information of numbers of defects connected region, area of each connected region and the total area through the method of analyzing defects characteristics. In the module of defect recognition, according to PCB's five common defects of short circuit, open circuit, burr, defects and voids, this paper classifies them according to their different characteristics of defects, designs corresponding algorithm to judge the type of defect, and prompts the defect information, all of these can conveniently help operators to have a statistic.