摘要
从X射线焊缝底片数字化图像缺陷识别的实际需求出发,针对焊缝缺陷图像边缘提取,进行不同边缘检测方法的对比研究,分析各类边缘检测算子边缘提取效果的特点及阈值的取值对边缘检测算子的边缘提取效果的影响,提出基于数学形态学的多方位结构元素边缘检测方法.实验表明:综合采用阈值优化的Canny算子和基于改进的形态学梯度算子及多方位结构元素的数学形态学边缘检测方法,能够获得完整全面的边缘信息,有利于对焊缝缺陷的分类及缺陷等级识别.
To meet the demand on digitalized image defect recognition on X-ray welding-line negatives, different edge detection methods are compared based on the image edge extraction. By analyzing the effect of edge detection operators and the impact of operator thresholding on edge extraction, the mathematical morphology based edge detection method is proposed for multidirectional structure elements. From the experiments, it is found that, with integration of threshold-optimized Canny operator and improved morphological gradient operator, complete and comprehensive edge information can be obtained for welding-line defect classification and rating.
出处
《中国工程机械学报》
2013年第3期267-271,共5页
Chinese Journal of Construction Machinery
基金
国家自然科学基金资助项目(51105064)
关键词
边缘检测
数学形态学
结构元素
焊缝底片
阈值
edge detection
mathematical morphology
structural element
welding-line negative
threshold value
