基于图像的缸体零件表面缺陷检测方法  

Surface defect detection method for cylinder body parts based on images

作　　者：谭亚雄 刘伟[1] 张海洋[1,2] 

机构地区：[1]中国科学院长春光学精密机械与物理研究所,吉林长春130033 [2]中国科学院大学,北京100049

出　　处：《传感器与微系统》2016年第4期137-140,共4页Transducer and Microsystem Technologies

基　　金：国家“863”高新技术发展计划资助项目(2011AA12A103)

摘　　要：针对人工抽样目测方法和超声波、电磁波等测量技术用于新型检测方法存在各式各样的缺点,为了实现在线高速检测汽车制动缸内壁是否存在缺陷,基于数字图像处理技术,提出了一种能很好满足要求的图像处理和缺陷检测方法。通过CCD相机采集零件需要检测表面的图像;用Matlab软件对图像进行必要的处理、对边缘进行提取;通过腐蚀和膨胀运算优化边缘,再采用连通域标记,通过计算缺陷区域处面积和周长两个物理量来判断缺陷类型和特征。结果表明:此方法能检测出缸体内表面缺陷,基本上能满足缺陷快速检测要求。Aiming at all kinds of faults existed in artificial sampling visual method and measurement technology such as ultrasonic wave,electromagnetic wave applied for new type detection methods,in order to realize high speed online detection on flawed in auto brake cylinder wall,propose an image processing and defect detection method well meet the requirements based on digital image processing technology. Collect images on surface of parts needed to detect by CCD camera; process image by Matlab software,and extract edge; optimize edge by etching and expansion operation,again use tag of connected domain,judge types and characteristics of defect by calculating two physical quantities which are defect area and perimeter. The results show that the method can detect the cylinder inner-surface defects and can satisfy the requirement of rapid detection of defects basically.

关 键 词：数字图像处理 边缘提取 缺陷识别 非接触性检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]