磁粉探伤缺陷识别自动化系统设计与开发
|
摘要
随着国际竞争的日趋激烈,各行各业对于产品质量与性能的要求不断提出更高的要求。磁粉探伤作为迄今为止对铁磁性材料表面及近表面伤痕检测最有效的无损检测手段,已经得到非常广泛的使用,但是现行使用的磁粉探伤设备始终摆脱不了采用人工目测进行最终缺陷识别的弊端,其主要不足表现在以下几个方面：一是检测速度慢,工作效率低；二是操作人员劳动强度大且工作内容单调重复,导致漏检率高；三是荧光磁粉探伤工作现场紫外光对长时间工作的人员造成比较严重的身体伤害。本课题基于上述缺陷与不足,设计开发一套荧光磁粉探伤自动化缺陷识别系统,实现利用计算机代替人眼完成工件伤痕的自动化识别。
文章首先分析了磁粉探伤缺陷识别自动化系统设计开发的相关技术发展与研究现状,指出当前技术存在的不足之处,采用模块化设计方法,构建了基于磁粉探伤技术与计算机视觉处理技术的系统总体结构与实现方案。通过对磁痕图像信息特征的分析,抓住目标信息与伪信息的特征差异,结合运用阂值图像分割与数学形态学处理原理,设计了基于Photoshop色彩范围选择功能与基于梯度特征差异的两种图像分割方法,实现了磁痕图像中目标信息与背景信息的准确分离；然后基于优化的区域生长理论设计了可疑连通域的快速提取算法,并抓住真伪伤痕特征差异对其进行了准确的判断识别；最后基于VC6.0编程工具与OpenCV计算机视觉库的运用,完成了整套方案的软件实现。
本磁粉探伤缺陷识别自动化系统基于CJW-1000磁粉探伤试验平台进行了调试与运行,结果表明系统运行稳定,效果良好,达到了预先的设计目标,为自动化荧光磁粉探伤设备研制工作的成熟与完善奠定了基础。
With the increasingly fierce international competition, a higher request to function and performance of products are brought forward in all walks of life. Magnetic detection, as the most effective and safe means of detecting the bruiser on and near the surface of magnetic materials so far, has been widely used. But the currently used magnetic detection equipment still can't get rid of the problem of finally recognizing the defects by using manual observation. The main shortcomings are as follows:First, the detection speed is slow with low efficiency; Second, big operating personnel labor intensity, along with monotonic and cyclic job contents, leads to high leak detection rate; Third, the UV-light on the job site poses sever health hazard to the operating personnel working long hours. To overcome the above shortcomings, a new automatic flaw recognition system of fluorescent magnetic detection is designed to automatically recognize the bruise by using computers instead of human eyes.
This paper first analyzes relevant technology development and research situation of the design of automatic flaw recognition system of magnetic detection, and points out the shortcomings of current technology. The modularizing design method is adopted, and an overall structure and its realization method are built based on the magnetic detection technology and computer vision processing technique. Through analyzing magnetic marks photos, capturing the difference between target information and false information, combined with threshold method and mathematical morphology treatment theory, two color image segmentation methods are designed based on gradient difference, and the target information and the background information are separated accurately. And then the single connected region is extracted efficiently based on the optimization of region growing algorithm, and the true and the false of the bruise is precisely judge based on the character differences between true and false flaw. Last, the whole scheme is accomplished based on the utilization of VC6.0 and OpenCV.
The automatic flaw recognition system of magnetic detection is debugged on the CJW-1000 magnetic detection experimental platform. The results demonstrates that the system runs steadily, the effects are good, which achieved the design targets, and established foundation for the maturity and perfection of the research works of automatic fluorescent magnetic detection equipment.
文章首先分析了磁粉探伤缺陷识别自动化系统设计开发的相关技术发展与研究现状,指出当前技术存在的不足之处,采用模块化设计方法,构建了基于磁粉探伤技术与计算机视觉处理技术的系统总体结构与实现方案。通过对磁痕图像信息特征的分析,抓住目标信息与伪信息的特征差异,结合运用阂值图像分割与数学形态学处理原理,设计了基于Photoshop色彩范围选择功能与基于梯度特征差异的两种图像分割方法,实现了磁痕图像中目标信息与背景信息的准确分离；然后基于优化的区域生长理论设计了可疑连通域的快速提取算法,并抓住真伪伤痕特征差异对其进行了准确的判断识别；最后基于VC6.0编程工具与OpenCV计算机视觉库的运用,完成了整套方案的软件实现。
本磁粉探伤缺陷识别自动化系统基于CJW-1000磁粉探伤试验平台进行了调试与运行,结果表明系统运行稳定,效果良好,达到了预先的设计目标,为自动化荧光磁粉探伤设备研制工作的成熟与完善奠定了基础。
With the increasingly fierce international competition, a higher request to function and performance of products are brought forward in all walks of life. Magnetic detection, as the most effective and safe means of detecting the bruiser on and near the surface of magnetic materials so far, has been widely used. But the currently used magnetic detection equipment still can't get rid of the problem of finally recognizing the defects by using manual observation. The main shortcomings are as follows:First, the detection speed is slow with low efficiency; Second, big operating personnel labor intensity, along with monotonic and cyclic job contents, leads to high leak detection rate; Third, the UV-light on the job site poses sever health hazard to the operating personnel working long hours. To overcome the above shortcomings, a new automatic flaw recognition system of fluorescent magnetic detection is designed to automatically recognize the bruise by using computers instead of human eyes.
This paper first analyzes relevant technology development and research situation of the design of automatic flaw recognition system of magnetic detection, and points out the shortcomings of current technology. The modularizing design method is adopted, and an overall structure and its realization method are built based on the magnetic detection technology and computer vision processing technique. Through analyzing magnetic marks photos, capturing the difference between target information and false information, combined with threshold method and mathematical morphology treatment theory, two color image segmentation methods are designed based on gradient difference, and the target information and the background information are separated accurately. And then the single connected region is extracted efficiently based on the optimization of region growing algorithm, and the true and the false of the bruise is precisely judge based on the character differences between true and false flaw. Last, the whole scheme is accomplished based on the utilization of VC6.0 and OpenCV.
The automatic flaw recognition system of magnetic detection is debugged on the CJW-1000 magnetic detection experimental platform. The results demonstrates that the system runs steadily, the effects are good, which achieved the design targets, and established foundation for the maturity and perfection of the research works of automatic fluorescent magnetic detection equipment.
引文
[1]叶代平,苏李广.磁粉检测[M].北京：机械工业出版社,2004
[2]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,1987
[3]李喜孟.无损检测[M].大连理工大学出版社.2008
[4]陈健生.磁粉探伤磁化技术与磁化设备[J].无损检测,24(7),2006：298～306
[5]徐振佩.铁路轮对荧光磁粉探伤缺陷识别系统研究[D].南京理工大学,2008
[6]陈健生,王海民,付洋.磁粉探伤20年回顾[J].无损检测,16(4),1998：91
[7]ANIL K. JAIN著,韩博译.数字图像处理基础[M].北京：清华大学出版社,2006
[8]Rafael C G, Richard E W. Digital image processing[M]. London:Prentice Hall,2006
[9]田岩,彭复员.数字图像处理与分析[M].武汉：华中科技大学出版社,2009
[10]陈渊宇.计算机在磁粉探伤机中的应用开发[D].南京理工大学,2010
[11]张杰.轮箍表面荧光磁粉探伤系统[D].安徽大学,2006
[12]吕作超.荧光磁粉探伤缺陷识别系统图像处理技术研究[D].南京理工大学,2007,(01)
[13]赵彦玲.基于图像技术的钢球表面缺陷分析与识别[D].哈尔滨理工大学,2008
[14]Milan Sonka, Vaclav Hlavac, Roger Boyle. Image Processing, Analysis and Machine Vision[M]. Thomson Learning and PT Press, USA.2003
[15]Hongwei Hao, Luming Li, Yuanlui Deng. Vision System Using Linear CCD Cameras in Fluorescent Magnetic Particle Inspection of Axles of Railway Wheel Sets[J]. Health Monitoring and Smart Nondestructive Evaluation of Structural and Biological System IV, edited by Tribikram Kundu, Proc.of SPIE.2005(5768):442～449
[16]A.S.Bakunov, A.Yu.Korolev, D.A.Kudryavtsev, V.P.Petrov. A Set of Magnetic Fluorescent Penetrant Inspection[J]. Russian Journal of Nondestructive Testing.2005 (3):170～174
[17]王恒迪,尚振东,马伟.轴承套圈磁粉探伤机的研制[J].轴承,32(4),2008：15～17
[18]周兆,白龙海,张泽彪,程洋.荧光渗透法无损检测的原理与运用[J].实验科学与技术,7(1),2009：50～53
[19]刘磊.全自动荧光磁粉检测系统的分析与改进[D].北京工业大学,2005
[20]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(3)：141～144
[21]吴海超,曾奇夫,查苏倩.荧光磁粉探伤裂纹智能识别图像处理研究[J].铁道技术监督,22(10),2010：77～82
[22]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(1)：7-10
[23]杨淑莹.VC++图像处理程序设计[M].清华大学出版社,2003
[24]何斌,马天予,王运坚等Visual C++数字图像处理[M].北京：人民邮电出版社,2005.
[25]章毓晋.图像处理与分析技术[M].北京：高等教育出版社,2008
[26]Gray B, Adrian K. Learning OpenCV[M]. Frankfort:O'Reilly Media,Inc.2009
[27]刘磊.全自动荧光磁粉检测系统的分析与改进[D].北京工业大学,2005
[28]李如玮.铁路轮对全自动荧光磁粉探伤系统的研究与实现[D].北京工业大学,2002
[29]白新伟.磁粉检测系统中图像分割的方法研究[D].长春理工大学,2009
[30]高庆吉,胡丹丹,牛国臣等.基于磁光图像的飞机铆钉缺陷识别[J].中国图像图形学报,12(12),2007：80-84
[31]Adrian S, Inna K, Eitan Y, et al. Recognition of motion-blurred images by use of the method of moments[J]. Applied Optics,41(11),2009:2164～2171
[32]Cheng H D. Color image segementation:advances and prospects[J]. Pattern Recognition, 34(21),2007:2259～2281
[33]Liu J Q, Yang Y H. Multiresolution color image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,16(7):689～700
[34]Min J, Bowyer K W. Improved range image segmentation by analyzing surface fit patterns[J]. Computer Vision and Image Understanding,97(2),2005:242～258
[35]Malik J. Normalized cuts and image segmentation[J], IEEE Tran sactions on Pattern Analysis and Machine Intelligence,22(8),2000:888～905
[36]杨阳,蒋先刚.基于图像分析的滚动轴承表面缺陷识别技术研究[J].华东交通大学学报,25(6),2008：41-46
[37]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),2010,34(6)：803-808
[38]张强,霍凯.轴承荧光磁粉探伤自动识别技术的研究[J].现代电子技术,2009,32(7)：107-110
[39]Bruce F,Jeff S著.谢君英,杜玲译Photoshop CS3 camera raw完全剖析[M].北京：人民邮电出版社,2009
[40]A.S.Bakunov, A.Yu.Korolev, D.A.Kudryavtsev, V.P.Petrov. A Set of Magnetic Fluorescent Penetrant Inspection[J]. Russian Journal of Nondestructive Testing.2005 (3):170～174
[41]张玉珍,王建宇,戴跃伟.基于自适应双阈值和主色率的足球视频镜头的分割[J].南京理工大学学报(自然科学版),2009,33(4)：432～437.
[42]崔屹.图像处理与分析-数学形态学的方法及应用[M].北京：科学出版社,2000
[43]李峰,徐诚,任国全,等.基于数学形态学的铁谱磨粒图像分割研究[J].南京理工大学学报(自然科学版),2005,29(1)：70～72
[44]陆建峰,杨静宇,叶玉坤.一个用于彩色肺癌细胞图像的分割算法[J].南京理工大学学报(自然科学版),2000,24(6)：481～485
[45]Robert D F, Theodore A T. Image quality of sparse-aperture designs for remote sensing[J].41(5),2008:1957～1968
[46]A B Meinel, Meinel M P. Inflatable membrand mirrors for optical passbandimagery[J]. 39(4),2009:541～550
[47]Vio R, Nagay, Tenorio L, et al. A simple but effient algorithm for multiple-image deblurring[J]. Astronomy &Astrophysics,416(5),2003:403～410
[48]Shi F Y, Chuang C F. A modified regulated morphological corner detector[J]. Pattern Recognition Letters,23(4),2005:931～937
[49]原培新,孙岩,陈波,纪革.图像处理在X射线胶片缺陷识别中的应用[J].CT理论与应用研究,16(1),2007：34～47
[50]Chen C S, Wu J L, Huang Y P. Theoretical aspects of vertically invariantbray-level morphological operators and their application on adaptive signal and image filtering[J]. IEEE Trans signal process,47(4),1999:1049～1060
[51]Laurent Itti,Carl Gold Christof Koch. Visual attention and target detection in cluttered natural scenes.Optical Engineering,40(9),2001:730～741
[52]Diego Santa-Cruz, T Ebrahimi. An analytical study of JPEG 2000 functionalities[J]. Proceedings of IEEE International Conference on Image Processing (ICIP-2000), Vancouver, BC.2(2),2000:49～52
[53]莫国影、左敦稳,朱笑笑.基于CCD图像的表面疲劳裂纹识别与长度计算[J].机械制造与研究,34(9),2007：55～56
[54]高原,胡绍海.复杂海平面下序列图像的快速检测[A].第十三届全国信号处理学术年会(CCSP-2007)论文集[C],2007：923～927
[55]牛照东,邹江威,王卫华,陈曾平.基于形状结构和梯度方向加权的Hausdorff距离图像匹配方法[A].第十四届全国信号处理学术年会(CCSP-2009)论文集[C],2009：566～570
[56]吴海滨,郑宏伟,李明琥,蒋锦涛.轮毂表面自动荧光磁粉探伤系统及其图像处理技术[J].无损检测,29(3),2007：128～131
[57]韩辉.数字化超声波探伤仪关键技术的研究[D].沈阳理工大学,2009
[58]李小泉.自动检测系统关键技术研究[D].武汉理工大学,2006
[59]周军,彭沛欣.自动磁粉探伤系统中的图像技术[J].仪器仪表学报,24(4)：2003：461-462
[60]刘瑞祯,于仕琪OpenCV教程-基础篇[M].北京：北京航空航天大学出版社,2007
[61]蒋先刚.数字图像模式识别工程软件设计[M].北京：中国水利水电出版社,2008
[62]沈昊.基于DSP图像处理的鸡蛋新鲜度实时无损检测研究[D].华中农业大学,2010
[63]陈胜勇.基于OpenCV的计算机视觉技术实现[M].北京：科学出版社,2008
[2]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,1987
[3]李喜孟.无损检测[M].大连理工大学出版社.2008
[4]陈健生.磁粉探伤磁化技术与磁化设备[J].无损检测,24(7),2006：298～306
[5]徐振佩.铁路轮对荧光磁粉探伤缺陷识别系统研究[D].南京理工大学,2008
[6]陈健生,王海民,付洋.磁粉探伤20年回顾[J].无损检测,16(4),1998：91
[7]ANIL K. JAIN著,韩博译.数字图像处理基础[M].北京：清华大学出版社,2006
[8]Rafael C G, Richard E W. Digital image processing[M]. London:Prentice Hall,2006
[9]田岩,彭复员.数字图像处理与分析[M].武汉：华中科技大学出版社,2009
[10]陈渊宇.计算机在磁粉探伤机中的应用开发[D].南京理工大学,2010
[11]张杰.轮箍表面荧光磁粉探伤系统[D].安徽大学,2006
[12]吕作超.荧光磁粉探伤缺陷识别系统图像处理技术研究[D].南京理工大学,2007,(01)
[13]赵彦玲.基于图像技术的钢球表面缺陷分析与识别[D].哈尔滨理工大学,2008
[14]Milan Sonka, Vaclav Hlavac, Roger Boyle. Image Processing, Analysis and Machine Vision[M]. Thomson Learning and PT Press, USA.2003
[15]Hongwei Hao, Luming Li, Yuanlui Deng. Vision System Using Linear CCD Cameras in Fluorescent Magnetic Particle Inspection of Axles of Railway Wheel Sets[J]. Health Monitoring and Smart Nondestructive Evaluation of Structural and Biological System IV, edited by Tribikram Kundu, Proc.of SPIE.2005(5768):442～449
[16]A.S.Bakunov, A.Yu.Korolev, D.A.Kudryavtsev, V.P.Petrov. A Set of Magnetic Fluorescent Penetrant Inspection[J]. Russian Journal of Nondestructive Testing.2005 (3):170～174
[17]王恒迪,尚振东,马伟.轴承套圈磁粉探伤机的研制[J].轴承,32(4),2008：15～17
[18]周兆,白龙海,张泽彪,程洋.荧光渗透法无损检测的原理与运用[J].实验科学与技术,7(1),2009：50～53
[19]刘磊.全自动荧光磁粉检测系统的分析与改进[D].北京工业大学,2005
[20]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(3)：141～144
[21]吴海超,曾奇夫,查苏倩.荧光磁粉探伤裂纹智能识别图像处理研究[J].铁道技术监督,22(10),2010：77～82
[22]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(1)：7-10
[23]杨淑莹.VC++图像处理程序设计[M].清华大学出版社,2003
[24]何斌,马天予,王运坚等Visual C++数字图像处理[M].北京：人民邮电出版社,2005.
[25]章毓晋.图像处理与分析技术[M].北京：高等教育出版社,2008
[26]Gray B, Adrian K. Learning OpenCV[M]. Frankfort:O'Reilly Media,Inc.2009
[27]刘磊.全自动荧光磁粉检测系统的分析与改进[D].北京工业大学,2005
[28]李如玮.铁路轮对全自动荧光磁粉探伤系统的研究与实现[D].北京工业大学,2002
[29]白新伟.磁粉检测系统中图像分割的方法研究[D].长春理工大学,2009
[30]高庆吉,胡丹丹,牛国臣等.基于磁光图像的飞机铆钉缺陷识别[J].中国图像图形学报,12(12),2007：80-84
[31]Adrian S, Inna K, Eitan Y, et al. Recognition of motion-blurred images by use of the method of moments[J]. Applied Optics,41(11),2009:2164～2171
[32]Cheng H D. Color image segementation:advances and prospects[J]. Pattern Recognition, 34(21),2007:2259～2281
[33]Liu J Q, Yang Y H. Multiresolution color image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,16(7):689～700
[34]Min J, Bowyer K W. Improved range image segmentation by analyzing surface fit patterns[J]. Computer Vision and Image Understanding,97(2),2005:242～258
[35]Malik J. Normalized cuts and image segmentation[J], IEEE Tran sactions on Pattern Analysis and Machine Intelligence,22(8),2000:888～905
[36]杨阳,蒋先刚.基于图像分析的滚动轴承表面缺陷识别技术研究[J].华东交通大学学报,25(6),2008：41-46
[37]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),2010,34(6)：803-808
[38]张强,霍凯.轴承荧光磁粉探伤自动识别技术的研究[J].现代电子技术,2009,32(7)：107-110
[39]Bruce F,Jeff S著.谢君英,杜玲译Photoshop CS3 camera raw完全剖析[M].北京：人民邮电出版社,2009
[40]A.S.Bakunov, A.Yu.Korolev, D.A.Kudryavtsev, V.P.Petrov. A Set of Magnetic Fluorescent Penetrant Inspection[J]. Russian Journal of Nondestructive Testing.2005 (3):170～174
[41]张玉珍,王建宇,戴跃伟.基于自适应双阈值和主色率的足球视频镜头的分割[J].南京理工大学学报(自然科学版),2009,33(4)：432～437.
[42]崔屹.图像处理与分析-数学形态学的方法及应用[M].北京：科学出版社,2000
[43]李峰,徐诚,任国全,等.基于数学形态学的铁谱磨粒图像分割研究[J].南京理工大学学报(自然科学版),2005,29(1)：70～72
[44]陆建峰,杨静宇,叶玉坤.一个用于彩色肺癌细胞图像的分割算法[J].南京理工大学学报(自然科学版),2000,24(6)：481～485
[45]Robert D F, Theodore A T. Image quality of sparse-aperture designs for remote sensing[J].41(5),2008:1957～1968
[46]A B Meinel, Meinel M P. Inflatable membrand mirrors for optical passbandimagery[J]. 39(4),2009:541～550
[47]Vio R, Nagay, Tenorio L, et al. A simple but effient algorithm for multiple-image deblurring[J]. Astronomy &Astrophysics,416(5),2003:403～410
[48]Shi F Y, Chuang C F. A modified regulated morphological corner detector[J]. Pattern Recognition Letters,23(4),2005:931～937
[49]原培新,孙岩,陈波,纪革.图像处理在X射线胶片缺陷识别中的应用[J].CT理论与应用研究,16(1),2007：34～47
[50]Chen C S, Wu J L, Huang Y P. Theoretical aspects of vertically invariantbray-level morphological operators and their application on adaptive signal and image filtering[J]. IEEE Trans signal process,47(4),1999:1049～1060
[51]Laurent Itti,Carl Gold Christof Koch. Visual attention and target detection in cluttered natural scenes.Optical Engineering,40(9),2001:730～741
[52]Diego Santa-Cruz, T Ebrahimi. An analytical study of JPEG 2000 functionalities[J]. Proceedings of IEEE International Conference on Image Processing (ICIP-2000), Vancouver, BC.2(2),2000:49～52
[53]莫国影、左敦稳,朱笑笑.基于CCD图像的表面疲劳裂纹识别与长度计算[J].机械制造与研究,34(9),2007：55～56
[54]高原,胡绍海.复杂海平面下序列图像的快速检测[A].第十三届全国信号处理学术年会(CCSP-2007)论文集[C],2007：923～927
[55]牛照东,邹江威,王卫华,陈曾平.基于形状结构和梯度方向加权的Hausdorff距离图像匹配方法[A].第十四届全国信号处理学术年会(CCSP-2009)论文集[C],2009：566～570
[56]吴海滨,郑宏伟,李明琥,蒋锦涛.轮毂表面自动荧光磁粉探伤系统及其图像处理技术[J].无损检测,29(3),2007：128～131
[57]韩辉.数字化超声波探伤仪关键技术的研究[D].沈阳理工大学,2009
[58]李小泉.自动检测系统关键技术研究[D].武汉理工大学,2006
[59]周军,彭沛欣.自动磁粉探伤系统中的图像技术[J].仪器仪表学报,24(4)：2003：461-462
[60]刘瑞祯,于仕琪OpenCV教程-基础篇[M].北京：北京航空航天大学出版社,2007
[61]蒋先刚.数字图像模式识别工程软件设计[M].北京：中国水利水电出版社,2008
[62]沈昊.基于DSP图像处理的鸡蛋新鲜度实时无损检测研究[D].华中农业大学,2010
[63]陈胜勇.基于OpenCV的计算机视觉技术实现[M].北京：科学出版社,2008