简介:Graycrosscorrelationmatchingtechniqueisadoptedtoextractcandidatematcheswithgraycrosscorrela-tioncoefficientslessthansomecertainrangeofmaximalcorrelationcoefficientcalledmulti-peakcandidatematches.Multi-peakcandidatesareextractedcorrespondingtothreeclosestfeaturepointsatfirst.Thecorrespondingmulti-peakcandidatematchesareusedtoconstructthemodelpolygon.Correspondenceisdeterminedbasedonthelocalgeometricrelationsbetweenthethreefeaturepointsandthemulti-peakcandidates.Thedisparitytestandtheglobalconsistencycheckoutareappliedtoeliminatetheremainingambiguousmatchesthatarenotremovedbythelocalgeometricrelationaltest.Experimentalresultsshowthattheproposedalgorithmisfeasibleandaccurate.