Research On The Detection Method Based On Image For Defect In Circuit Board

With the rapid development of electronic technology and the large demand forprinted circuit board, printed circuit board technology is more and more advancetowards the trend of high-precision,high-density and multi-layered，the circuit boardmanufacturer becomes has great concern on the PCB inspection technology. Combinedwith industrial support plans of Jiangxi province “High-Accuracy Measurement SystemBased on Image”, In this paper, the technology of registration between standard circuitboard image and detecting circuit board image, and defects automatic detection andcategory recognition of circuit board image is deeply studied. The main work andachievements are as follows:1) Obtain standard circuit board image from the Gerber file is studied. The methodtakes into account the Gerber file is the standard data file of the circuitt board,acquiringX-Y coordinate data of the circuit board lines and pads by deeply parsing the instructionin Gerber file, and generating standard circuit board image for defects automaticdetection and recognition of circuit board image.2) The preprocessing method of detecting circuit board image. For the uniqueproduction of the circuit board and the inevitable introduction of CCD camera to obtainthe detecting circuit board image. Firstly, gaussian noise and pepper noise is reduced byfast median filter.Then, the circuit board lines and pads of circuit board image isseparated from the background by the method of maximum classses square. Finally, theedge profile of the circuit board lines and pads is extracted by canny operator, could beused for circuit board image registration.3) The registration of standard circuit board image and detecting circuit boardimage is studied, a registration algotithm based on the range of prohibting wiring boxand the center of positioning hole is proposed. Firstly, we extract parameters ofprohibting wiring box and positioning hole of the standard circuit board image byGerber file. Secondly, detecting circuit board image is angle-Correctted based on leastsquares estimate algorithm and the range of positioning hole is automatic searched.Then, we obtain the parameters of the positioning hole of detecting circuit board imageby least squares estimate algorithm. Finally, the registration of two circuit board images accomplished by parameters of the positioning hole.Experimental results show that thealgorithm can effectively finishing the registration of circuit board images.4) Defects automatic detection and recognition of circuit board image is studied, anew algorithm is proposed. Firstly, we realize the defects automatic detection of circuitboard image by difference image method and mathematical morphology. Then,according to the characteristics of common defects, synthesizing area method,connected domain method and defect edge neighbor method, achieve the automaticidentification of defect types. Finally, The experimental results of the real PCB imagesof simulation of defect detection demonstrate that the algorithm is fast, accurate andeffective.