简介:MostexistingapplicationsofcentroidalVoronoitessellations(CVTs)lackconsiderationofthelengthoftheclusterboundaries.Inthispaperweproposeanewmodelandalgorithmstoproducesegmentationswhichwouldminimizethetotalenergy—asumoftheclassicCVTenergyandtheweightedlengthofclusterboundaries.TodistinguishitwiththeclassicCVTs,wecallitanEdge-WeightedCVT(EWCVT).TheconceptofEWCVTisexpectedtobuildamathematicalbaseforallCVTrelateddataclassificationswithrequirementofsmoothnessoftheclusterboundaries.TheEWCVTmethodiseasyinimplementation,fastincomputation,andnaturalforanynumberofclusters.
简介:CentroidalVoronoitessellations(CVTs)havebecomeausefultoolinmanyapplicationsrangingfromgeometricmodeling,imageanddataanalysis,andnumericalpartialdifferentialequations,toproblemsinphysics,astrophysics,chemistry,andbiology.Inthispaper,webrieflyreviewtheCVTconceptandafewofitsgeneralizationsandwell-knownproperties.WethenpresentanoverviewofrecentadvancesinbothmathematicalandcomputationalstudiesandinpracticalapplicationsofCVTs.Wheneverpossible,wepointoutsomeoutstandingissuesthatstillneedinvestigating.
简介:Wepresentanovelalgorithmforadaptivetriangularmeshcoarsening.Thealgorithmhastwostages.First,theinputtriangularmeshisrefinedbyiterativelyapplyingtheadaptivesubdivisionoperatorthatperformsaso-calledred-greensplit.Second,therefinedmeshissimplifiedbyaclusteringalgorithmbasedoncentroidalVoronoitessellations(CVTs).TheaccuracyandgoodqualityoftheoutputtriangularmeshareachievedbycombiningadaptivesubdivisionandtheCVTstechnique.Testresultsshowedthemeshcoarseningschemetoberobustandeffective.Examplesareshownthatvalidatethemethod.
简介:Wetackletheproblemofconstructing2DcentroidalVoronoitessellationswithconstraintsthroughanefficientandrobustconstructionofboundedVoronoidiagrams,thepseudo-dualoftheconstrainedDelaunaytriangulation.WeexploitthefactthatthecellsoftheboundedVoronoidiagramcanbeobtainedbyclippingtheordinaryonesagainsttheconstrainedDelaunayedges.Theclippingitselfisefficientlycomputedbyidentifyingforeachconstrainededgethe(connected)setoftriangleswhosedualVoronoiverticesarehiddenbytheconstraint.TheresultingconstructionisamenabletoLloydrelaxationsoastoobtainacentroidaltessellationwithconstraints.
简介:Thispaperconsidershowtouseagroupofrobotstosenseandcontroladiffusionprocess.Thediffusionprocessismodeledbyapartialdifferentialequation(PDE),whichisabothspatiallyandtemporallyvariantsystem.Therobotscanserveasmobilesensors,actuators,orboth.CentroidalVoronoiTessellationsbasedcoveragecontrolalgorithmisproposedforthecooperativesensingtask.Forthediffusioncontrolproblem,thispaperconsiderssprayingcontrolviaagroupofnetworkedmobilerobotsequippedwithchemicalneutralizers,knownassmartmobilesprayersoractuators,inadomainofinteresthavingstaticmeshsensornetworkforconcentrationsensing.ThispaperalsointroducestheinformationsharingandconsensusstrategywhenusingcentroidalVoronoitessellationsalgorithmtocontroladiffusionprocess.Theinformationissharednotonlyonwheretospraybutalsoonhowmuchtosprayamongthemobileactuators.BenefitsfromusingCVTandinformationconsensusseekingforsensingandcontrolofadiffusionprocessaredemonstratedinsimulationresults.
简介:研究D-Cchang等人引进的五个区域Hardy空间,刻划这些空间的原子分解和对偶空间,揭示了这些空间的内在联系。