Study On Detection Method Of Ceramic Substrate Defect Based On Machine Vision

Ceramic substrate is a ceramic plate.Its production process is to first stamping and then sintering molding.Because substrate material is relatively soft,it has a degree of uncertainty of the product which affects its applications when substrate material is pressed.The defect detection of the ceramic substrate mainly detects its appearance quality including cracks,spots,defects,etc.At present,ceramic substrate defect detection is mainly rely on artificial to complete.This way has a low detection speed,and often results in test error because of the visual fatigue of workers.Using machine vision technology for industrial measurement can not only avoid the disadvantages of manual detecting action,but also can improve the accuracy and the efficiency of detection,and reduce the production cost.This paper studies the ceramic substrate defect detection system based on machine vision technology.The system obtains the image information of the ceramic substrate through CCD camera,removes image noise by pre-processing in the PC,then corrects the tilt of the ceramic substrate.In the correction process of ceramic substrate tilt,firstly,the image edge should be extracted based on sobel operator method;secondly,refine the image edge based on RobertS operator.After getting the image edge,by comparing three conventional tilt correction algorithm—the Hough transform,Radon transform,rotating projection,we choose the improved-algorithm based on the rotation projection method to measure the rotation angle of the image,finally using bilinear interpolation realize of the ceramic substrate image tilt correction.After correction of the image,we detect the defect of ceramic substrate.Defects of ceramic substrate generally include defects,burst,porosity,impurities,indentation,and no nicks aspects.For these,this paper studies the ceramic substrate defects detection system based on machine vision technology.Specific research work includes several aspects:In the detection process of ceramic substrate defect,using threshold segmentation to extract the region of ceramic,determining whether the defect by comparing the area;At the time of the holes and other defects were detected,comparison of the defect area from the surrounding gray background difference;When detecting dimpling,using Gaussian low pass filter to make the image smooth;When detecting ceramic substrate without scoring,using template matching method image and subtraction,find out the area without scoring.