摘要
焊缝缺陷检测对于焊接结构件的可靠性具有重要意义。使用机器视觉系统替代人工检测,不但节省劳动力,而且检测精度更高。由于焊缝缺陷图像常存在噪声、对比度不强的现象,以及焊缝缺陷形态具有多样性的特征,使得焊缝缺陷难以准确检测出来,容易造成漏检或误检。使用各向异性的Curvelet变换技术,处理焊缝缺陷图像后,焊缝缺陷边缘更清晰。根据Curvelet系数的特征,对不同尺度层分别计算自适应阈值的值,再使用该阈值根据半软阈值函数的方法对Curvelet系数进行处理,经Curvelet反变换后图像能够保存边缘细节的同时,滤除焊缝缺陷图像中的噪声,使焊缝缺陷边缘十分清晰,从而提高焊缝缺陷的检测精度。
Weld defect detection has a great significance for the reliability of welded structures. It is not only laboursaving, but also has higher accuracy that machine vision system replaces artificial inspection. The weld defect images often have noise and the low contrast, what's more, the form characteristics of weld defect is diverse. It is hard to detect weld defect accurately and causes weld defect missed or false detection easily. Weld defect edges are more clearly by anisotropic Curvelet transform technology. According to the characteristics of Curvelet coefficient, adaptive threshold value is respectively calculated to every scales layer. And then, Curvelet coefficient is processed by the threshold and the method of semi-soft threshold function. By inverse Curvelet transform, the edge details of defect in image can be preserved and the noise of the weld defect image is eliminated. The method makes the edge of weld defect become clear so as to improve the detection accuracy of weld defect.
出处
《焊接技术》
2015年第6期57-61,0,共5页
Welding Technology
基金
基于机器视觉流体微小流量测量系统研究(GKZY201102)
移动破碎站液压自动控制系统开发(GK201209)
基于电力线室内定位技术(2010XJKRL011)
