基于机器视觉的轻触开关引脚缺陷检测算法研究  

Study on the defect inspection algorithm of touch switch pins based on machine vision

作　　者：蒋建东[1] 陈培余 童一珏 WILLEM Hoogmoed 

机构地区：[1]浙江工业大学特种装备制造与先进加工技术教育部重点实验室,浙江杭州310014 [2]瓦赫宁根大学农场技术组

出　　处：《浙江工业大学学报》2015年第1期30-33,38,共5页Journal of Zhejiang University of Technology

基　　金：国家自然科学基金资助项目(51375456);浙江省高校中青年学科带头人攀登计划项目(pd2013021)

摘　　要：针对轻触开关在生产过程中外观检测的要求,进行了基于机器视觉的轻触开关引脚外观图像检测技术研究.通过CCD获取图像,采用滤波、阈值分割等预处理提取了轻触开关的二值图像,并采用轮廓提取,根据最小面积包围法获取图像绕中心旋转角度的方法对图像进行矫正,对旋转后的图像提取4个引脚区域的ROI区域.对提取引脚区域的黑色像素面积进行统计,并与设定的标准面积阈值进行对比判断进行轻触开关引脚外观缺陷的检测.实验结果表明:引脚歪脚的识别率为98%,引脚合格的误判率为3%,综合准确率大于97%,所提出的检测方法及算法能准确检测轻触开关的引脚缺陷,满足轻触开关外观检测的需求.This paper studies the appearance detection technology of touch switch pins based on machine vision,according to the appearance detecting requirements of touch switch in the production process.Firstly,the requested images are gained through CCD sensor,and then get the binary image by filtering and threshold segmentation pretreatment.Then extract the contour of the image and rectify the image by the angle of rotation around the center area according the smallest enclosing method,after that get the four regions of interest（ROI）of the switch pins.Last get statistics of the pin black pixels area and compare with the standard area threshold to judge the appearance defects of the touch switch pins.The experimental results show that the method and algorithm can precisely detect the pin defects and meet the requirement of touch switch appearance detection with recognition rate of crooked pin 98%,rate of qualified error 3%,accuracy rate greater than 97%.

关 键 词：轻触开关 机器视觉 外观检测 缺陷 ROI 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]