欢迎来到天天文库
浏览记录
ID:58140213
大小:1.19 MB
页数:5页
时间:2020-04-24
《基于先验知识和纹理方向性的车标定位方法-论文.pdf》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、第31卷第5期计算机应用与软件V01.31No.52014年5月ComputerApplicationsandSoftwareMav2014基于先验知识和纹理方向性的车标定位方法王建任明武(南京理工大学计算机科学与技术学院江苏南京210094)摘要基于先验知识,根据车牌与车标的相对位置关系,首先对车标进行粗定位,得到包含车标图案的矩形粗定位区域。为了利用车标背景散热片的纹理方向特征,引入纹理方向测度的概念。根据纹理测度将散热片纹理分为两大类:类水平纹理和垂直纹理。在精定位阶段,对于不同的纹理类型采用不同的定位方法:对于类水平纹理,采用动态模板定位边界,对复杂水平纹理和复杂网状纹理具有良
2、好的定位效果;对于垂直纹理,引入边缘颗粒度的概念,既有效地消除非车标区域的噪声点,又完整地保留车标本身的信息。实验结果表明,该方法鲁棒性好,能够快速、准确地定位车标位置,平均成功定位率为97%,在测试环境下平均定位时间为22ms,定位效率和定位成功率能满足实时系统的需要。关键词车标定位纹理方向特征边缘密度中图分类号TP391.41文献标识码ADOI:10.3969/j.issn.1000·386x.2014.05.042VEHICLE.LoGOLoCATINGMETHoDBASEDoNPRIORKNOWLEDGEANDTEXTUREDIRECTIoNALITYWangJianRenMi
3、ngwu(SchoolofCompu~rScienceandTechnology,Nang西eofScienceandTechnology,Na210094,Jiangsu,China)AbstractBasedonthepriorknowledgeandaccordingtotherelativepositionrelationshipofvehiclelicenceplateandvehicle—logo,flintwemakeroughpositioningofthevehiclelogoandgetarectangularroughlocationregioncontainin
4、gthevehicle—logoimage.Inordertoutilisethetexturedirectionalfeatureofthevehicle—logobackgroundcoolingglls,weintroducetheconceptoftexturedirectionmeasure.Accordingtothetexturemeasure,thetexturesoftheglleayegroupedintotwocategories:thehorizontal—liketextureandtheverticaltexture.Inthestageofaccurate
5、location,weadoptdiferentlocationalgorithmsinlightofdiferenttexturecategoriesrespectively.Forhorizontal—liketexture,weusedynamictemplatestolocatetheboundary,thisachievessoundlocationeffectsforsophisticatedhorizontaltextureandcomplicatedmeshtexture;Forverticaltexture,weintroducetheconceptofedgegra
6、nularity,iteffectivelyremovesthenoisespointsoutsidethevehiclelogoregionandpreservesentireinformationofthevehicle-logoitsown.Experimentalresultsshowthatthemethodinthispaperhasgoodrobustness,itcanfastandaccuratelylocatethepositionofvehicle·logo,theaveragesuccessfullocatingratereaches97%,andtheaver
7、agelocatingtimeis22msintestingenvironment.Thislocationefficiencyandaccuracycanmeetthedemandofreal—timesystem.KeywordsVehicle—logolocationTexturedirectionalfeatureEdgedensity用的瓶颈⋯,并且车标一旦定位错误,后期识别工作将毫无意0引言义。现有的车标定位算法包括:基于Adabo
此文档下载收益归作者所有