机器视觉在厚壁钢管端面缺陷检测中的应用

Application of Machine Vision in Detecting the Defects of the End Face of Thick-wall Steel Tube

针对目前人工检测厚壁钢管端面缺陷存在的效率低、速度慢,且还会出现错检、漏检等问题,提出一种基于机器视觉的方法,实现对厚壁钢管端面缺陷的检测及分类。首先单独提取钢管倒角区域,利用最小二乘法对内外倒角包含的轮廓圆进行拟合,并根据欧式距离来判断倒角是否出现偏心的情况;其次提取钢管端面区域,并通过Otsu算法分割出缺陷区域,计算各联通域的特征描述并组成新的特征向量,使用支持向量机来判断缺陷类型。研究结果表明:该方法能准确检测出厚壁钢管的倒角是否偏心、端面是否存在各类的缺陷,且准确率达到96.7%,对一钢管端面的判断时间不超过100 ms,相比人工目测速度有明显的提高。

Aiming at the problems of low efficiency and slow speed of manual detection of steel pipe end face as well as wrong and missing inspection,a method based on machine vision was proposed to detect and classify the defects of thick-walled steel pipe end face. Firstly, separating the region of the internal and external chamfering of the steel pipe,and using the least-squares method to fit the contour circle included in the internal and external chamfering which can be estimated whether the chamfering of steel pipe is deviated or not. Secondly, using the method of Otsu to separate the defects area from the end face of the steel pipe,and after extracting the features of the connected domain as well as making up a new eigenvector,the defects lastly could be detected and classified by SVM. The results show that this method can correctly detect the chamfering of the thick-walled steel pipe and whether there are any defects on the end face. The accuracy rate is up to 96.7 %,and the detection time is less than 100 ms,which has an obvious improvement compared to manual inspection.