基于模糊聚类和形态学的轮胎断面特征提取-江苏大学.PDF
文本预览下载声明
2012年9月 第33卷 第5期
Sept.2012 Vol.33 No.5
doi:10.3969/j.issn.1671-7775.2012.05.004
基于模糊聚类和形态学的轮胎断面特征提取
王国林,周树仁,李军强
(江苏大学 汽车与交通工程学院,江苏 镇江212013)
摘要:针对轮胎断面图像中胶料之间灰度相差小,边界模糊性强,传统的边缘提取方法难以实现断
面特征提取的问题,提出了1种将多层次模糊增强、模糊 C均值聚类以及形态学等方法有机结合,
用于轮胎断面特征提取的算法.该算法先将钢丝区域去除并对图像整体进行灰度变换,然后通过多
层次模糊增强,增加胶料之间的对比度,再运用模糊C均值聚类分割出各胶料区域,并对各区域进
行形态学处理并提取边界,最后将各区域边界叠加,加入钢丝区域,得到轮胎断面的特征.对不同光
照条件下的轮胎断面图像进行试验.结果表明:该算法可以解决轮胎断面图像中胶料之间灰度差小
的问题,并有效提取了轮胎断面中的胶料边界和钢丝特征,且对光照条件具有一定的适应性.
关键词:轮胎断面;多层次模糊增强;模糊C均值聚类;形态学
中图分类号:U463.341 文献标志码:A 文章编号:1671-7775(2012)05-0513-05
Featureextractionoftiresectionbasedonfuzzyclustering
andmorphology
WangGuolin,ZhouShuren,LiJunqiang
(SchoolofAutomotiveandTrafficEngineering,JiangsuUniversity,Zhenjiang,Jiangsu212013,China)
Abstract:Thefeaturesoftiresectionweredifficulttobeextractedbythetraditionaledgedetectionme
thodsbecauseofthesmalldifferenceofgrayvaluebetweenrubbersinthetiresectionimageandthefuzzy
boundaries.Analgorithmwasproposedtoextractthefeaturesoftiresectionbycombiningmultilevel
fuzzyenhancement
,fuzzyCmeansclusteringandmorphology.Theregionsofsteelwireswereremovedto
transformthegrayvaluesofimage.Themultilevelfuzzyenhancementwasusedtoincreasethecontrast
betweenrubbers.TherubberregionswereeffectivelysegmentedbyfuzzyCmeansclustering,andthe
boundarieswereextractedbymorphology.Thefeaturesoftiresectionwereobtainedbystackingthe
boundaryofeachregionandaddingsteelwires.Experimentsoftiresectionimagewereconductedunder
differentlightconditions.Theresultsshowthatthealgorithmcaneffectivelysolvetheproblemofsmall
grayvaluedifferencebetweenrubbersintheimageoftiresectionandextractthefeaturesoftiresection
显示全部