Abstract. Local feature approaches to vision geometry and object recognition are based on selecting and matching sparse sets of visually salient image points, known as ‘keypoints ’ or ‘points of interest’. Their performance depends critically on the ac
DetectingKeypointswithStablePosition,Orientation
andScaleunderIlluminationChanges
BillTriggs
GRAVIR-CNRS-INRIA,655avenuedel’Europe,38330Montbonnot,France.Bill.Triggs@inrialpes.fr doc.guandang.net
Abstract.Localfeatureapproachestovisiongeometryandobjectrecognitionarebasedonselectingandmatchingsparsesetsofvisuallysalientimagepoints,knownas‘keypoints’or‘pointsofinterest’.Theirperformancedependscriticallyontheaccuracyandreliabilitywithwhichcorrespondingkeypointscanbefoundinsubsequentimages.Amongthemanyexistingkeypointselectioncriteria,thepopularF¨orstner-Harrisapproachexplicitlytargetsgeometricstability,de ningkeypointstobepointsthathavelocallymaximalself-matchingprecisionundertranslationalleastsquarestemplatematching.However,manyapplicationsrequirestabilityinorientationandscaleaswellasinposition.Detectingtranslationalkey-pointsandverifyingorientation/scalebehaviourposthocissuboptimal,andcanbemisleadingwhendifferentmotionvariablesinteract.Wegiveamoreprincipledformulation,basedonextendingtheF¨orstner-Harrisapproachtogeneralmotionmodelsandrobusttemplatematching.Wealsoincorporateasimplelocalappear-ancemodeltoensuregoodresistancetothemostcommonilluminationvariations.Weillustratetheresultingmethodsandquantifytheirperformanceontestimages.Keywords:keypoint,pointofinterest,cornerdetection,featurebasedvision,F¨orstner-Harrisdetector,templatematching,visiongeometry,objectrecognition.1Introduction
Local-feature-basedapproacheshaveprovensuccessfulinmanyvisionproblems,in-cludingscenereconstruction[16,5],imageindexingandobjectrecognition[20,21,32,33,23,24,25].Thebasicideaisthatfocusingattentiononcomparativelysparsesetsofespeciallysalientimagepoints—usuallycalledkeypointsorpointsofinterest—bothsavescomputation(asmostoftheimageisdiscarded)andimprovesrobustness(astherearemanysimple,redundantlocalcuesratherthanafewpowerfulbutcomplexanddeli-categlobalones)[37].However,localmethodsmustbeableto nd‘thesame’keypointsagaininotherimages,andtheirperformancedependscriticallyonthereliabilityandaccuracywithwhichexactlycorrespondingpointscanbefound.Manyapproachestokeypointdetectionexist,including‘corners’[2,17,38,28,4],parametricimagemodels
[3,31,1],localenergy/phasecongruency[27,29,30,18],andmorphology[35,19].OneofthemostpopularisthatdevelopedbyF¨orstner&G¨ulch[7,9]andHarris&Stephens
[15]followingearlierworkbyHannah[14]andMoravec[26].Thisapproachbrings
开机停在detecting array_IT/计算机_专业资料。电脑开机停在 detecting array... 把硬盘线拔了试试。。看看能不能过这个一步,如果过了这步就是硬盘。如果不 ...
开机显示 DETECTING IDE DRIVES Detecting IDE drives 的意识是说正在检测 IDE 驱动,而你进不去证明你的 IDE 硬件有问题了,建议检查一下 IDE 营建。 先换个数据...
“Listening between the lines” means detecting the implied meaning, drawing inferences about the speaker’s intention from what is heard and what is not ...
Detecting SYN Flooding Attacks_IT/计算机_专业资料 暂无评价0人阅读0次下载举报文档 Detecting SYN Flooding Attacks_IT/计算机_专业资料。一种SYN Flood攻击...
Detecting putative orthologs Summary: We developed an algorithm that improves upon the common procedure of taking reciprocal best blast hits (rbh) in the ...
detecting faces in images_教育学/心理学_人文社科_专业资料。关于人脸识别的简介Detecting faces in images: a survey 2014/11/19 引用率 ? 1159 Citations ?...
2009DETECTING REAL LIFE ANGER_英语学习_外语学习_教育专区。DETECTING REAL LIFE ANGER Felix Burkhardt, *Tim Polzehl, Joachim Stegmann, *Florian Metze ...
方海泉论文报告Detecting Novel Associations in Large Data Sets_企业管理_经管营销_专业资料。Detecting Novel Associations in Large Data Sets David N. Reshef ...
detecting communities in large networks_工学_高等教育_教育专区。Detecting communities in large networks A. Capocci1 , V.D.P. Servedio12 , G. Calda...
Detecting Unusual Activity in Video_天文/地理_自然科学_专业资料 暂无评价0人阅读0次下载举报文档 Detecting Unusual Activity in Video_天文/地理_自然科学_...
热门文档