资源描述：
《Combining color and shape information for illumination-viewpoint invariant object recognition .pdf》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
1、IEEETRANSACTIONSONIMAGEPROCESSING,VOL.15,NO.1,JANUARY20061CombiningColorandShapeInformationforIllumination-ViewpointInvariantObjectRecognitionAristeidisDiplaros,StudentMember,IEEE,TheoGevers,Member,IEEE,andIoannisPatras,Member,IEEEAbstract—Inthispaper,weproposeanewsch
2、emethatmergesthecomputationalcomplexityisexponentialfornontrivialim-color-andshape-invariantinformationforobjectrecognition.Toages.Therefore,inthispaper,wefocusonappearance-basedobtainrobustnessagainstphotometricchanges,color-invariantobjectrecognition.derivativesarec
3、omputedfirst.ColorinvarianceisanimportantMostoftheworkonappearance-basedobjectrecognitionaspectofanyobjectrecognitionscheme,ascolorchangescon-siderablywiththevariationinillumination,objectpose,andbasedonshapeinformationisbymatchingsetsofshapeimagecameraviewpoint.Thesec
4、olorinvariantderivativesarethenfeatures(e.g.,edges,corners,andlines)betweenamodelandusedtoobtainsimilarityinvariantshapedescriptors.Shapeatargetimage.Infact,theprojectiveinvarianceofcrossra-invarianceisequallyimportantas,underachangeincameratiosanditsgeneralizationtoc
5、rossratiosofareasoftrianglesviewpointandobjectpose,theshapeofarigidobjectundergoesandvolumesoftetrahedrahasbeenusedforviewpoint-invariantaperspectiveprojectionontheimageplane.Then,thecolorandshapeinvariantsarecombinedinamultidimensionalcolor-shapeobjectrecognitionands
6、ignificantprogresshasbeenachievedcontextwhichissubsequentlyusedasanindex.Astheindexing[1].Othershapeinvariantsarecomputedbasedonmoments,schememakesuseofacolor-shapeinvariantcontext,itprovidesFouriertransformcoefficients,edgecurvature,andarclengthahigh-discriminativeinfo
7、rmationcuerobustagainstvarying[2],[3].Unfortunately,shapefeaturesarerarelyadequateforimagingconditions.Thematchingfunctionofthecolor-shapediscriminatoryrecognitionof3-Dobjectsfromarbitraryview-contextallowsforfastrecognition,eveninthepresenceofobjectocclusionandclutte
8、ring.Fromtheexperimentalresults,itispoints.Shape-basedobjectrecognitionisofteninadequatees-shownthatthemethodrecognizesrigid
