[1]陆宝春,李建文,陈吉朋,等.荧光磁粉探伤自动缺陷识别方法研究[J].南京理工大学学报(自然科学版),2010,(06):803-808.
 LU Bao-chun,LI Jian-wen,CHEN Ji-peng,et al.Automatic Flaw Recognition Method of Fluorescent Magnetic Detection[J].Journal of Nanjing University of Science and Technology,2010,(06):803-808.
点击复制

荧光磁粉探伤自动缺陷识别方法研究
分享到:

《南京理工大学学报》(自然科学版)[ISSN:1005-9830/CN:32-1397/N]

卷:
期数:
2010年06期
页码:
803-808
栏目:
出版日期:
2010-12-31

文章信息/Info

Title:
Automatic Flaw Recognition Method of Fluorescent Magnetic Detection
作者:
陆宝春;李建文;陈吉朋;王婧;李纯健;汤海昌;
南京理工大学机械工程学院
Author(s):
LU Bao-chun1LI Jian-wen1CHEN Ji-peng1WANG Jing1LI Chun-jian2TANG Hai-chang2
1.School of Mechanical Engineering,NUST,Nanjing 210094,China;2.Shengjieda NDT Device Manufacturing Co.,Ltd.,Yancheng 224300,China
关键词:
荧光磁粉探伤 缺陷识别 图像分割 图像形态学 区域生长
Keywords:
fluorescent magnetic detection flaw recognition image segmentation image morphology regional growing
分类号:
TP391.41
摘要:
针对荧光磁粉探伤的人工观察识别过程中工作量大、效率低、准确性低及健康危害严重等问题,提出一种新的自动化图像处理与缺陷识别方法。通过分析磁痕图像,设计了一种基于Photoshop中"色彩范围"选择功能的彩色图像分割算法;基于图像形态学提出图像去噪与还原算法;在区域生长理论的基础上进行相关优化,实现单个连通区域的高效提取;基于真伪信号的特征差异设计了缺陷识别方法。以样本图片为例,在集成了OpenCV的VC环境下进行了算法调试,结果显示:彩色图像分割模块滤除了原始图像中由于紫外灯光照射形成的蓝紫色区域和附着磁粉形成的暗绿色区域等干扰信号;开运算后图像中的斑点噪声已经完全被去除,部分细小区域在开运算过程中被断开,导致图像失真;通过闭运算图像得到了还原,图像特征指标保持良好;在识别效率显著改善的同时,识别方法的通用性与准确性也都得到了明显提高。
Abstract:
To overcome the problems of heavy workload,low efficiency,low accuracy and serious health hazard of the process of manual observation in fluorescent magnetic detection,a new automatic image processing and flaw recognition method is proposed.Through analyzing magnetic mark photos,a color image segmentation method is designed based on the choice function of "color range" in Photoshop.The image denoising and restoration algorithm is given based on an image morphology.The single connected region is extracted efficiently through the optimization of region growing algorithm.The flaw recognition method is designed based on the character differences between true and false signals.Taking the sample image for example,the algorithm is debugged in the VC environment integrated with OpenCV.The results show: the interference signal of the blue-purple region formed by ultraviolet light irradiation and the dark green region formed by attachment magnetic powder in the origin image is flitted by the color image segmentation model;the speckle noise is removed after open operation,image distortion is caused by disconnecting some smell regions in the process of open operation;the image is reduced by close operation,the character targets of the image are kept well;with the improvement of the recognition efficiency,the universality and accuracy of the recognition method is improved.

参考文献/References:

[ 1] 中国机械工程学会无损检测学会. 磁粉探伤[M ].北京: 机械工业出版社, 2005.

[ 2] 李喜孟. 无损检测[M ]. 大连: 大连理工大学出版社, 2001.
[ 3 ] 彭飞, 朱晓军, 朱志洁. 荧光磁粉探伤自动检测及图像处理系统研究[ J] . 船海工程, 2009, 38 ( 3 ):141- 144.
[ 4 ] 彭沛欣, 周军, 鲍志强. 荧光磁粉无损检测自动化系统的实现[ J]. 河海大学常州分校学报, 2003,17( 1): 7- 10.
[ 5] Rafael C G, Richard E W. D ig ita l im age process ing[M ]. London, Eng land: Prentice H a l,l 2006.
[ 6] G ray B, Adr ian K. Learning OpenCV [M ]. Frankfurt,Germ any: O. Re illyM edia, Inc, 2009.
[ 7] Bruce F, Jeff S. Photoshop CS3 camera raw 完全剖析[M ]. 谢君英, 杜玲, 译. 北京: 人民邮电出版社, 2009.
[ 8] M ilan S, Vac lavH, Roger B. Im age process ing, ana-lysis and m ach ine v ision [M ]. New Jersey, USA:Thom son Learn ing and PT Press, 2003.
[ 9] BakunovA S, Koro lev A Y, Kudryav tsev D A. A seto fm agne tic fluo rescent penetrant inspection[ J] . RussianJournal o fNondestructive Testing, 2005, 41( 3):170- 174.
[ 10] 孙日明. 用于医学图像分割的区域生长方法[ J].大连交通大学学报, 2010, 31( 2): 91- 94.[ 11] 杨淑莹. VC ++ 图像处理程序设计[M ]. 北京: 清华大学出版社, 2003.

相似文献/References:

[1]陆宝春,李建文,王婧,等.基于特征差异性的荧光磁粉探伤图像分割算法[J].南京理工大学学报(自然科学版),2011,(06):797.
 LU Bao-chun,LI Jian-wen,WANG Jing,et al.Feature Difference Based Image Segmentation Algorithm for Fluorescent Magnetic Detection[J].Journal of Nanjing University of Science and Technology,2011,(06):797.

备注/Memo

备注/Memo:
基金项目: 国家科技型中小企业技术创新基金( 09C26213203712); 国家大学生创新实验计划( 101022802)作者简介: 陆宝春( 1965- ), 男, 博士, 教授, 主要研究方向: 制造系统检测控制、诊断与维护, E-m a il: lbcnust@ s-ina. com。
更新日期/Last Update: 2012-11-02