Abstract: Thepointclouddataregistrationisoneofthekeytechnicalaspectsofthreedimensionalreconstructionfornoncooperativespacecraft.Tosolvethepairwiseregistrationissueofsparsepointcloud,andalignthemultipleviewpointcloudaccurately,anoptimizationregistrationmethodwaspresented.AnovelpointcloudsimplificationalgorithmusinguniformsamplingwasproposedbasedonKDTreeandtheuniformrelationofthethresholdparametersinthe4PCS(4PointsCongruentSets)algorithmwasestablishedviathedensityofthepointcloud.Byusingthemisregistrationcorrectionmethodbasedonattitude,themisregistrationproblemofthesymmetrypointcloudwassolved.Theresultsshowthattheproposedalgorithmcaneffectivelyachievegoodalignmentsofthesparsepointcloudofthesatellite,besides,thestabilityandthesuccessrateofthealgorithmisalsoimproved.

Key words: noncooperativespacecraft, multipleview, pointclouddata, registration, 4pointscongruentsets