基于形态学的焊接图像缺陷检测方法研究  

Algorithm for Welding Defect Detection of Ray Images Based on Mathematical Morphology

作　　者：马云[1] 王贺[2] 张晓光[1] 胡晓钦[1] 张涛[1] 

机构地区：[1]中国矿业大学机电工程学院,江苏徐州221008 [2]上海交通大学电子信息与电气工程学院,上海200240

出　　处：《计算机测量与控制》2010年第5期1008-1010,1013,共4页Computer Measurement &Control

基　　金：中国博士后科学基金资助项目(20060390277);江苏省工业高科技研究计划资助项目(BG2007013);江苏省"六大人才高峰"基金资助项目(06-E-52)

摘　　要：射线焊接缺陷的准确检测是保证焊接构件质量的关键,针对焊接生产中采集的图像存在对比度低、噪声干扰及边界模糊等特点,提出了一种基于灰度形态学的射线焊接图像的缺陷检测方法;首先,在对原始射线焊接图像滤波和增强的基础上,根据对焊接图像列灰度的分析运用阈值法提取出焊道;然后,选取了多尺度多结构元素,采用灰度形态学方法对焊接图像进行缺陷边缘检测,并二值化处理后提取出焊接图像中的缺陷;试验结果表明,相比传统的边缘检测算法,该方法能有效提取图像中的焊接缺陷边缘,且其连续性和完整性较好。In order to insure the quality of welding, the accurate defects detection in ray images of welding is important. In consider of characteristics of welding images such as low contrast, noise and fuzzy borders, an improved algorithm based on mathematical morphology for defect detection of ray images is proposed. Firstly, on the basis of pretreatment of original weld image, welding seam is extracted from the background; and then, an improved mathematical morphology method with the multi--scale and multi--elements is used to extract the welding defect edge. Experimental results show that the method can effectively extract a clearer edge of weld defects from welding images with nice continuity and integrity, compared to the traditional edge detection algorithms.

关 键 词：缺陷检测 数学形态学 焊接图像 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]