文献类型：期刊文献

中文题名：工业CT图像的缺陷检测研究

英文题名：Defect Detection Research of Industrial CT Images

作者：刘晶[1]

机构：[1]华东理工大学机械与动力工程学院,上海200237

年份：2020

期号：9

起止页码：118

中文期刊名：机械设计与制造

外文期刊名：Machinery Design & Manufacture

收录：CSTPCD;;北大核心:【北大核心2017】；

基金：中央高校基本科研业务费专项资金(222201714016)。

语种：中文

中文关键词：工业CT;缺陷检测;阈值分割;亚像素边缘检测

外文关键词：Industrial CT;Defect Defection;Threshold Segmentation;Subpixel Edge Defection

摘要：铸造过程由于其自身生产工艺的特点,常常会产生缩孔和气孔等缺陷,这些内部缺陷会对零件质量产生不好的影响,也会缩短零件寿命,因此需要准确地识别出这些内部缺陷。通过工业CT设备对铸件进行扫描,可以获得了一系列工业CT切片图像。为了快速提取工业CT图像的孔类缺陷,首先使用二维Otsu自适应阈值算法进行阈值分割,以区分工业CT图像中的物体与背景,然后通过Sobel算子得到图像的初始边缘轮廓,再基于拉格朗日插值法进行亚像素边缘检测。实验表明,该方法可以有效地识别铸件工业CT图像中的缩孔和气孔缺陷。
The casting process often causes shrinkage and porosity because of its characteristics.These internal defects have bad effect on the quality of parts and can short the life of the casting.So these defects need be identified accurately.In this paper,the casting is scanned by industrial Computed Tomography(CT)equipment.A series of CT section images are got.Hole defects are extracted quickly.First,threshold segmentation is done by two dimensional Ostu adaptive threshold algorithm.Thus the object and the background in the industrial CT image are distinguished.Then the initial contour edge is extracted by Sobel operator.Finally,subpixel edge is detected based on Lagrange interpolation.Experiments show hole defects of the industrial CT image can be detectedeffectively.