资源描述:
《基于svm和决策树的自然图像低能见度天气现象识别》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、JournalofImageandSignalProcessing图像与信号处理,2016,5(4),155-165
PublishedOnlineOctober2016inHans.http://www.hanspub.org/journal/jisp
http://dx.doi.org/10.12677/jisp.2016.54018LowVisibilityWeatherRecognitionviaSVM
andDecisionTreeinSingleImageGuanleiXu*,XiaotongWang,LiminShao,LijiaZh
2、ou,XiaogangXuOceanDepartmentofDalianNavyAcademy,DalianLiaoningReceived:Sep.17th,2016;accepted:Oct.4th,2016;published:Oct.7th,2016Copyright©2016byauthorsandHansPublishersInc.ThisworkislicensedundertheCreativeCommonsAttributionInternationalLicense(CCBY).
http://creativecommons.o
3、rg/licenses/by/4.0/OpenAccessAbstractThevisibilityinaregionnotonlyreflectsthequalityoftheatmosphericenvironment,butalsohas
closerelationshipwithpeople’slife.Ingeneral,lowvisibilityweatheraffectspeople’seconomic
development,sothereal-timeobservationoflowvisibilityisofmuchsignif
4、ication.Thereasonof
lowvisibilityiscloselyassociatedwithmeteorologicalconditions.Thelowvisibilityweatherphe-
nomenamainlycontainrain,snow,fog,etc.Thispaperproposesarecognitionmethodwhichis
basedonlowvisibilityweatherphenomenonbymeansoftheinfluenceoflowvisibilityweather
phenome
5、nonontheimageinformationsuchastheimagecontrast,saturationandbrightnessthat
canbeemployedfortrainingandclassification.Weestablishaclassificationdecisiontreeaccord-
ingtothedistancebetweenthedifferentcategoriesintheprocessoftrainingandbuildingsupport
vectormachine(SVM)classifier
6、forthedecisiontree.Itcanclassifythelowvisibilityweatherim-
ageautomaticallyandintelligently.Throughtestingahugeamountofimagesdownloadedfrom
theinternet,theexperimentalresultsshowthatweatherimagemeanrecognitionrateisover70%.
Afteradoptingthevotingschemeviadistributedrecognition
7、,thefinallowvisibilityweatherrec-
ognitionrateismorethan95%.KeywordsLowVisibilityWeatherImage,SupportVectorMachine(SVM),DecisionTree,Training基于SVM和决策树的自然图像低能见度天气现象识别徐冠雷*,王孝通,邵利民,周立佳,徐晓刚*通讯作者。文章引用:徐冠雷,王孝通,邵利民,周立佳,徐晓刚.基于SVM和决策树的自然图像低能见度天气现象识别[J].图
像与信号处理,2016,5(4):155-165.http:/
8、/dx.doi.org/10.12677/jisp.2016.54018徐冠雷等海军大连舰艇学院军事海洋系,辽宁大连收稿日