欢迎来到天天文库
浏览记录
ID:11331781
大小:521.00 KB
页数:12页
时间:2018-07-11
《matlab图像处理 外文翻译 外文文献 英文文献 基于视觉的矿井救援机器人场景识别》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、附录A英文原文ScenerecognitionforminerescuerobotlocalizationbasedonvisionCUIYi-an(崔益安),CAIZi-xing(蔡自兴),WANGLu(王璐)Abstract:AnewscenerecognitionsystemwaspresentedbasedonfuzzylogicandhiddenMarkovmodel(HMM)thatcanbeappliedinminerescuerobotlocalizationduringemergencies.Thesystemusesmonocularcameratoacquireomni-
2、directionalimagesofthemineenvironmentwheretherobotlocates.Byadoptingcenter-surrounddifferencemethod,thesalientlocalimageregionsareextractedfromtheimagesasnaturallandmarks.TheselandmarksareorganizedbyusingHMMtorepresentthescenewheretherobotis,andfuzzylogicstrategyisusedtomatchthesceneandlandmark.Byth
3、isway,thelocalizationproblem,whichisthescenerecognitionprobleminthesystem,canbeconvertedintotheevaluationproblemofHMM.Thecontributionsoftheseskillsmakethesystemhavetheabilitytodealwithchangesinscale,2Drotationandviewpoint.Theresultsofexperimentsalsoprovethatthesystemhashigherratioofrecognitionandloc
4、alizationinbothstaticanddynamicmineenvironments.Keywords:robotlocation;scenerecognition;salientimage;matchingstrategy;fuzzylogic;hiddenMarkovmodel1IntroductionSearchandrescueindisasterareainthedomainofrobotisaburgeoningandchallengingsubject[1].Minerescuerobotwasdevelopedtoenterminesduringemergencies
5、tolocatepossibleescaperoutesforthosetrappedinsideanddeterminewhetheritissafeforhumantoenterornot.Localizationisafundamentalprobleminthisfield.Localizationmethodsbasedoncameracanbemainlyclassifiedintogeometric,topologicalorhybridones[2].Withitsfeasibilityandeffectiveness,scenerecognitionbecomesoneoft
6、heimportanttechnologiesoftopologicallocalization.Currentlymostscenerecognitionmethodsarebasedonglobalimagefeaturesandhavetwodistinctstages:trainingofflineandmatchingonline.Duringthetrainingstage,robotcollectstheimagesoftheenvironmentwhereitworksandprocessestheimagestoextractglobalfeaturesthatreprese
7、ntthescene.Someapproacheswereusedtoanalyzethedata-setofimagedirectlyandsomeprimaryfeatureswerefound,suchasthePCAmethod[3].However,thePCAmethodisnoteffectiveindistinguishingtheclassesoffeatures.Another
此文档下载收益归作者所有