期刊文献+

最小误差准则与脉冲耦合神经网络的裂缝检测 被引量:19

Detection of crack defect based on minimum error and pulse coupled neural networks
下载PDF
导出
摘要 表面裂缝检测能够有效判断混凝土桥梁出现的结构性危险。但裂缝特征的多样性、桥梁表面污点引起的图像噪声以及不均匀照明引起的灰度不均等给裂缝检测带来极大的困难。为能够在复杂背景下检测裂缝,分析裂缝图像特征,由脉冲耦合神经网络(pulse coupled neural networks,PCNN)的运行特征和神经元的状态变化分析简化PCNN模型,将简化PCNN模型用于裂缝图像的分割,根据最小误差准则判断PCNN迭代的终止条件,实现了PCNN的裂缝图像自动分割。由圆形度与扁度结合计算区域特征,去除分割后的各种干扰,实现表面裂缝的有效检测。通过敏感度和特异性计算绘制ROC(receiver operating charac-teristics)曲线,比较不同分割方法的曲线特性以评估算法,对实际裂缝图像的处理结果表明了该方法对裂缝图像检测的有效性。 Surface crack detection can effectively judge structure dangers of concrete bridge.But,crack detection becomes very difficult because of variety of crack characters,image noise caused by bridge surface blots and uneven gray scale caused by asymmetric illumination.In order to detect cracks in complicated background,crack image character is analyzed;PCNN model is simplified through analyzing of its running characters and the state change of nerve cells.Crack image is segmented using the simplified PCNN model;the iterative stop condition of the PCNN model is judged with the rule of minimum error,and PCNN crack image segmentation is carried out automatically.The region characters are calculated according to the degrees of flatness and roundness,the interferences after segmentation are removed,and the surface crack effective detection is achieved.ROC curves are drawn using sensitivity and specificity,and the curve characteristics of different detection methods are compared to evaluate the algorithm.Experiment results using the real images of bridge surface show that the proposed crack detection method is effective.
出处 《仪器仪表学报》 EI CAS CSCD 北大核心 2012年第3期637-642,共6页 Chinese Journal of Scientific Instrument
基金 长江学者和创新团队发展计划(RT0705)资助项目
关键词 裂缝检测 脉冲耦合神经网络 最小误差准则 ROC曲线 crack detection pulse coupled neural network minimum error criterion ROC curve
  • 相关文献

参考文献15

  • 1CHENG H D,SHI X J,GLAZIER C.Real-time imagethresholding based on sample space reduction and inter-polation approach[J].Journal of Computing in Civil En-gineering,2003,17(4):264-272.
  • 2HUANG Y X,XU B G.Automatic inspection of pave-ment cracking distress[J].Journal of Electronic Ima-ging,2006,15(1):1-5.
  • 3SUBIRATS P,DUMOULIN J,LEGEAY V.Automationof pavement surface crack detection with a matched filte-ring to define the mother wavelet function used[C].14thEuropean Signal Processing Conference,EUSIPCO 2006,3037-3040.
  • 4IKHLAS A Q,SARA P R.PCA Based algorithm for un-supervised bridge crack detection[J].Advances in Engi-neering Software,2006,37:771-778.
  • 5WANG Z B,MA Y D,CHENG F Y,Review of pulse-coupled neural networks[J].Image and Vision Compu-ting,2010,28(1):5-13.
  • 6张淼,韩光,钟映春,韦丽兴.基于脉冲耦合神经网络的光驱物镜导线品质检验方法的研究[J].仪器仪表学报,2011,32(7):1570-1577. 被引量:2
  • 7葛继,王耀南,张辉,周博文.基于改进型PCNN的智能灯检机研究[J].仪器仪表学报,2009,30(9):1866-1873. 被引量:21
  • 8JOHNSON L J,PADGETT M L.PCNN models and ap-plications[J].IEEE Transactions on Neural Networks,1999,2(3):480-498.
  • 9马义德,戴若兰,李廉.一种基于脉冲耦合神经网络和图像熵的自动图像分割方法[J].通信学报,2002,23(1):46-51. 被引量:145
  • 10KAPUR N,SAHOO P K,WONG A K C.A new methodfor gray-level picture thresholding using the entropy of thehistogram[J].Computer Graphics,Vision and ImageProcessing,1985,29(3):273-285.

二级参考文献32

  • 1张兵,卢焕章.序列图像中运动点目标轨迹检测算法研究[J].电子学报,2004,32(9):1524-1526. 被引量:15
  • 2张飞,李承芳,史丽娜,孙哓玮.复杂背景下运动点目标的检测算法[J].光学技术,2005,31(1):55-57. 被引量:17
  • 3赵静,方新林,孟力.静脉输液中不溶性微粒危害的预防措施[J].现代预防医学,2005,32(9):1213-1213. 被引量:34
  • 4李杨果,王耀南,王威.基于机器视觉的大输液智能灯检机研究[J].光电工程,2006,33(11):69-74. 被引量:30
  • 5MURESAN R C. Pattern recognition using pulse-coupled neural networks and discrete fourier transforms[J]. Neurocomputing, 2003,51:487-493.
  • 6KARVONEN J A. Baltic sea ice SAR segmentation and classification using modified pulse-coupled neural networks[J]. IEEE Trans. Geosci Remote Sensing, 2004,42(7) 1566-1574.
  • 7郑筱蔓等.中华人民共和国药典[M].北京:化学工业出版社,2005:743.
  • 8THOMAS A D H, RODD M G, HOLT J D, et cl. Real-time industrial visual inspection: a review[J]. Neurobiology of Learning and Memory. 1995,1(2):139-158.
  • 9ZHAO Y L, ZHANG ZH CH, GAO ZH M. A simple and workable moving objects segmentation method[C]. International Symposium Electronics in Marine, ELMAR2034, Zadar, Croatia, 2004(6):585-590.
  • 10HONG S, FENG SH. A real-time algorithm for moving objects detection in video images[C]. Proceedings of the 5th world Congress on Intelligent Control and Automation, Hangzhou, China,2004(5):4108-4111.

共引文献190

同被引文献190

引证文献19

二级引证文献212

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部