基于图像的铸造缺陷类型识别  

Type Identification of Casting Defects Based on Image

作　　者：刘晶[1] LIU Jing(School of Mechanical and Power Engineering,East China University of Science and Technology,Shanghai 200237,China)

机构地区：[1]华东理工大学机械与动力工程学院,上海200237

出　　处：《机械工程师》2021年第11期13-15,共3页Mechanical Engineer

基　　金：中央高校基本科研业务费专项资金(222201714016)。

摘　　要：针对存在缺陷的铸件,首先通过工业CT设备扫描铸件,得到一系列工业CT图像,通过二值化方法对图像目标和背景进行区分,然后进行轮廓跟踪、提取。在轮廓提取过程中,粗提取得到像素级坐标,精提取得到亚像素级坐标,这样可以更加精确地确定缺陷边缘数据。将轮廓数据进行排序后,最后通过计算圆形度和狭长度,确定其形状是更倾向于圆形或长条纹形,进而对缺陷类型进行识别。实验表明该方法可以有效地识别出孔类缺陷和裂纹类缺陷。For defective castings,a series of industrial CT images can be obtained by scanning.For these images,the object and the background of these images can be discriminated by image binaryzation.Then contour tracking which includes rough selection and accurate extraction is used to recognize all the contour of the image.In the process of contour extraction,the pixel level coordinates are obtained by rough extraction,and the subpixel level coordinates are obtained by fine extraction,so that the defect edge data can be determined more accurately.After the contour data are sorted in the same direction,it is determined that the shape is circular or long-striped by calculating the circularity and elongnatedness.Then the defect type is identified.Experiments show that the method can recognize hole and crack defect availably.

关 键 词：铸造缺陷 孔缺陷 裂纹缺陷 工业CT图像 缺陷类型识别 

分 类 号：TP319.7[自动化与计算机技术—计算机软件与理论]