Research On Algorithm Of PCB Defect Detection Based On Image Analysis

In the modern electronic equipment and electronic products, printed circuit board PCB is the connected wires among the various electronic components in various fields obtained large scale application. Detection is one of the important links in printed circuit board in printed circuit board manufacturing process,it is the necessary means in quality assurance and the production cost controlled.As the PCB's output is more and more big, the wiring density and accuracy are increasingly high, traditional detecting way as artificial visual and needle type detection can't satisfy the modern PCB production needs. Many enterprises need to achieve on-line detection based on machine vision of PCB defects detection system. This paper not only assures the quality of defects detection,but also puts forward some methods to reduce testing time and space complexity, the method of research content and major innovations as follows:First in image collection on the separate block collection, block processing methods, not only save the processing time, still can save them a lot of buffer space.For PCB gray image enhancement processing, with the improved based second-order gradient image of sharpening method, not only make the image of gray contrast greatly increased, and effectively highlighted the edge information, in addition to the noise. Experimental results show that the proposed based second-order gradient enhancement method is effective method of image enhancement, simple calculation, image enhancement while filter except near the edge of image noise, very suitable for online processing.PCB image and sample alignment for defect detection, the key problem is how to make the image and sample rapid, accurate alignment. This article USES the alignment way is through treat retrieval image and sample image comparing calculation, ask out of both position transformation parameters, by using the genetic algorithm to find the best transform parameter, make two images of minimum self-conflicts. Due to genetic algorithm using global parallel search, so can quickly search to global optimal value, is applied to the PCB image alignment can reduce time complexity, greatly enhancing detection system of real-time.Established based on the edge of the sample data structure through comparison of characteristic parameters search some blemish and defect classification, overcome traditional template matching algorithm to precise positioning and calculation of large amount of data faults. System USES Visual c + + 6.0 programming, realized the fine cutting positioning, rapid and accurate detection of defects tagging defects and the defect classification, and other functions, experimental results show that the detection a circuit board for no more than 20 seconds, accuracy of 99%.