| 基于DR图像多重快速分割的铁路货车铸件缺陷提取方法 |
| |
| 引用本文： | 曾理,洪达平,刘玲慧.基于DR图像多重快速分割的铁路货车铸件缺陷提取方法[J].中国铁道科学,2012,33(1):103-107. |
| |
| 作者姓名： | 曾理 洪达平 刘玲慧 |
| |
| 作者单位： | 1. 重庆大学数学与统计学院,重庆401331;重庆大学光电技术及系统教育部重点实验室ICT研究中心,重庆400044
2. 重庆大学光电技术及系统教育部重点实验室ICT研究中心,重庆,400044 |
| |
| 基金项目： | 国家自然科学基金资助项目(60972104) |
| |
| 摘 要： | 针对铁路货车铸件DR(Digital Radiography)图像数据量大、背景亮度不均匀以及对DR图像中铸件薄壁区域的缺陷提取效果不理想和基于C-V模型的分割方法存在计算量大的问题,研究基于DR图像多重快速分割的铸件缺陷提取方法.先利用OTSU双阈值分割算法对铸件DR图像进行分割,然后应用区域生长技术得到铸件薄壁区域的图像；对该图像进行剥皮和归一化线性灰度拉伸处理后,再次利用OTSU双阈值分割算法和区域生长技术对图像进行分割,从中提取出含有和可能含有铸件缺陷的局部图像；对可能含有铸件缺陷的亮背景区域图像用C-V模型再次进行分割,最终提取出所有铸件缺陷的图像.
|
| 关 键 词： | 铁路货车 铸件缺陷 图像识别 C-V模型 OTSU阈值分割 |
Defect Extraction Method for the Casting of Railway Freight Car Based on Multiple and Fast Segmentation of DR Image |
| |
| ZENG Li
,
HONG Daping
,
LIU Linghui.Defect Extraction Method for the Casting of Railway Freight Car Based on Multiple and Fast Segmentation of DR Image[J].China Railway Science,2012,33(1):103-107. |
| |
| Authors: | ZENG Li HONG Daping LIU Linghui |
| |
| Institution: | 1.College of Mathematics and Statistics,Chongqing University,Chongqing 401331,China; 2.ICT Research Center,Key Laboratory of Optoelectronic Technology and System of the Ministry of Education, Chongqing University,Chongqing 400044,China) |
| |
| Abstract: | In order to overcome the difficulties caused by the properties of DR(Digital Radiography) image of casting on railway freight car,such as the huge amount of data,background luminance inhomogeneity,the poor performance of segmentation results for thin wall region of DR image and the expensive computation of C-V model,in this paper,we study a defect extraction method for casting based on multiple and fast segmentation of DR image.Firstly,the OTSU dual-threshold method is used to segment the DR image of casting,and then we obtain the images of casting thin wall region by using region growing method.Secondly,after the images are treated by peeling and normalized linear gray stretching,OTSU dual-threshold method and region growing method are used again to segment the images to obtain the local images which may contain casting defects.Finally,C-V model is used to segment the casting defects in the bright local region of DR image,and all the defects in casting can be detected. |
| |
| Keywords: | Railway freight car Casting defect Image recognition C-V model OTSU threshold segmentation |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
