形态学滤波窄裂纹检测与目标识别 |
| |
引用本文: | 李仁兴,;张毅,;柏连发,;陈钱,;顾国华.形态学滤波窄裂纹检测与目标识别[J].无损检测,2009(10):796-799. |
| |
作者姓名: | 李仁兴 ;张毅 ;柏连发 ;陈钱 ;顾国华 |
| |
作者单位: | [1]江苏技术师范学院,常州213001; [2]南京理工大学,南京210014 |
| |
摘 要: | 荧光磁粉探伤是工件表面缺陷的一种非接触式检测手段,传统的基于人工视觉检测裂纹的方法耗人力、耗时、不精确、花费高、可靠性无法保证。现代工业检测技术要求工件表面缺陷检测自动完成,而工件表面状况、真伪裂纹缺陷、工况条件等使得现有的检测识别方法难以满足工件表面缺陷自动检测识别的需要。分析了工件表面荧光磁粉图像特征及裂纹缺陷特征;研究了基于分块阈值的数学形态学梯度算子图像边缘检测算法;根据裂纹缺陷的长宽比、圆形度等特征,设计了基于Fisher线性判别方法的工件裂纹缺陷识别方法。以此为基础的荧光磁粉探伤工件裂纹缺陷自动检测识别技术,应用于火车轮轴检测线实时检测,裂纹缺陷的有效检出率达90%。
|
关 键 词: | 磁粉探伤 自动检测 目标识别 图像分割 |
Morphological Filtering Slightness Crack Detection and Objects Identification |
| |
Affiliation: | LI Ren-Xing , ZHANG Yi , BAI Lian-Fa , CHEN Qian, GU Guo-Hua (1. Jiangsu Techers University of Technolojy, Changzhou 213001, China; 2. Nanjing University of Science and Technology, Nanjing 210014, China) |
| |
Abstract: | Magnetic powder detection is an important method for work-piece superficial crack detection. Traditional magnetic powder crack detection is manpower consuming, time consuming, high expenses, low precision, and fallibility. Modem industrial detection technology requires work-piece crack auto-detection. Because of exterior status, veritable or feigned crack ohiect, site condition, etc. , existing method can not successfully auto- detect and identify work-piece cracks. Fluorescent magnetic powder image and crack image characteristics are analyzed, morphological grads arithmetic operators image fringe detection based on sub--area threshold is studied, crack identification arithmetic based on Fisher linear discrimination is designed according to long-width ratio and round shape degree character. Cart-wheel-axis crack detection line equipped with this auto-detection and identification technology got an efficient crack detection ratio as high as 90%. |
| |
Keywords: | Magnetic powder crack detection Automatic detection Target identification Image segmentation |
本文献已被 维普 等数据库收录! |
|