Study On Algorithm Of PCB Quality Detection

Printed Circuit Board (PCB) plays an important role in modern electronic equipment and electronic products. It is an information carrier that integrates all kinds of electronic components and is widely applied to many fields. Along with the constant development of technology and continuous progress of industry, electronic products tend to become lighter, thinner, shorter and smaller. This makes PCB manufacture technology developing towards high-density. The PCB quality affects the capability of products directly. At present PCB quality test works are mostly completed by manual eyeballing. As the subjective factors'effects, missing and mistaken in test often happen. So it is urgent needed to take online automatic optical test into the PCB industry.Firstly, this paper introduces present situation and future development of PCB quality detection technology. Then simply presents the composing of PCB detection system and analyzes the character and characteristic of the equipments in the system. We research on the structure of PCB detection system which is integrated by computer visual technology, accurate machine technology, auto-control technology and image process technology. And we introduce elementary principle and method of digital image process and some mathematics method related to our task.This paper mainly discuss denoising and binarization of the test PCB, and the recognition of the defects. It proposes a novel method called fuzzy switching adaptive median filter to remove impulse noise. This method firstly judge whether a pixel was corrupted by impulse noise using fuzzy rules, followed by dealing with possible corrupted pixels using a new adaptive median filter. This method avoids the disadvantage of tradition median filter, which blurred the image when remove the noise, and effect of noise removal is much better than that of tradition median filter. We proposed a pertinency method to binarize the test image. This method employs wavelet to remove the noise in the gray histogram of test image firstly, and then uses the prior information of the standard image to find the threshold to binarize the test image. This paper also proposed an available method to remove the pseudo defects in the direct subtracted image, and recognize and classify the defects.By the experiment, simulation result proved: The accurate quality detection on PCB using proposed PCB detection algorithm have credible availability and real-time performance and fulfils demands of the PCB detection system.