欢迎来到天天文库
浏览记录
ID:28482190
大小:75.04 KB
页数:15页
时间:2018-12-10
《厚钢管x射线图像中焊缝区域的检测[w》由会员上传分享,免费在线阅读,更多相关内容在应用文档-天天文库。
1、厚钢管X射线图像中焊缝区域的检测关键字:钢管,射线图,图像,区域,区域的,检测厚钢管X射线图像中焊缝区域的检测本文为Word文档,感谢你的关注! 摘要:由于传统焊缝区域检测算法难以准确提取模糊和对比度低的厚钢管焊缝区域,提出一种新的基于鲁棒PCA模型的焊缝区域检测算法,该算法能克服传统方法的不足,并能准确提取焊缝区域.首先,收集一序列X射线图像,并对其进行空域对齐及亮度归一化预处理.其次,计算得到系列图像的背景图像,并将背景图像与待测试X射线图像张成一个观测矩阵.最后,使用鲁棒PCA算法对观测矩阵进行低秩与稀疏分解,测试图像中的不均匀强
2、度及噪声被消除,焊缝区域被凸显出来,通过全局阈值可将焊缝区域较好地分割出来.实验结果表明,该算法能较大地消除厚钢管X射线图像中噪声及不均匀强度分布带来的干扰、同时增强模糊的焊缝边缘及对比度低的区域,相比传统焊缝区域检测算法,具有更高的检测灵敏度(0.952)和精度(0.989),能更好地将模糊和对比度低的焊缝区域完整检测出来. 关键词:厚钢管;X-ray图像;焊缝区域;边缘检测;图像预处理 TP391文献标志码:A DetectionofWeldRegionsinX-rayImagesofThickSteelPipes CHENB
3、enzhi1,FANGZhihong2,XIAYong2,ZHANGLing3,LANShouren1,WANGLisheng1 (1.SchoolofElectronicInformationandElectricalEngineering,ShanghaiJiao TongUniversity,Shanghai200240,China;2.ResearchInstitute,BaoshanIronSteelCo, Shanghai201900,China;3.SteelBarsDivision,BaoshanIronSteelC
4、o,Shanghai201900,China) Abstract:Sincetraditionaldetectionalgorithmsofweldingseamareahavedifficultiesinaccuratelyextractingthefuzzyandlow-contrastweldingareasintheX-rayimagesofthicksteelpipes,thispaperproposedanovelrobustdetectionmethodofweldseamregionbasedontherobustPCA
5、model.Theproposedtechniquecanovercometheshortcomingsofthetraditionalmethods,andcanaccuratelyextracttheweldregions.Firstly,asequenceofX-rayimageswerecollected,andtheirspatialalignmentandbrightnessnormalizationwerecarriedout.Then,aseriesofbackgroundimageswereobtained,andthe
6、sepreprocessedimagesandatestX-rayimagewerecombinedtoformanobservationmatrix.TherobustPCAwasthenusedtodecomposetheobservationmatrixintoalow-rankandsparseimage.Astheunevenintensityandnoisearegreatlyeliminatedinthetestimages,theweldregionofthetestimageishighlightedinthecorre
7、spondingsparseimage,andcanbewellsegmentedbyaglobalthreshold.ThetestresultsshowthattheunevenbrightnessdistributionandnoisefromX-rayimagesofthicksteelpipesarelargelyeliminated,andtheweldseamedgesandlowcontrastareasarealsoenhanced.Comparedwiththetraditionalweldingareadetecti
8、onmethods,theproposedalgorithmcanbetterdetectthefuzzyandlow-contrastweldingregionswithahigherdet
此文档下载收益归作者所有