简介:Thispaperpresentsanewsplinecalledarcspline.Ittakestwoarcsconnectingadjacentpointsamongasampledseries,withknowntangentalofeachsampledpoint.Thecalculationfortheparallel.thecurvelengthandthesegment-ingpointcoordinatesaresimple,thusmakingmapsymbolgenerationsimplified.When'arclet'occurs,somespecialprocessingmustbetaken.
简介:Thispaperintroducesanoveltechniqueforobjectdetectionusinggeneticalgorithmsandmorphologicalprocessing.Themethodemploysakindofobjectorientedstructureelement,whichisderivedbygeneticalgorithms.Thepopulationofmorphologicalfiltersisiterativelyevaluatedaccordingtoastatisticalperformanceindexcorrespondingtoobjectextractionability,andevolvesintoanoptimalstructuringelementusingtheevolutionprinciplesofgeneticsearch.Experimentalresultsofroadextractionfromhighresolutionsatelliteimagesarepresentedtoillustratethemeritandfeasibilityoftheproposedmethod.
简介:Ithasbeendebatedthatwhichofrasterandvectoristhemosteffective,fromtheviewpointsofstorageefficiency,processingefficiency,andabilitytocapturetheessenceofgeographicspaceforanapplication.Theremaybenoclearanswertothisdebate.Itseemsappropriatetonotethatthemaindifferencebetweenthemrelatestothedegreetowhichthespatialentitiesofinterestareexplicitinthedata. Ingeneral,rasterremotesensingimageallowstheinvestigatortomeasureandmonitorsurfaceelectromagneticwavevariationinwhichmostobjectsandthespatialinterrelationshipsbetweenthemareimplicit.Inpractice,therearemanypoorlyunderstoodtradeoffsincouplingremotesensingdatatoexistingthematicmaps.Mapsusepointsandlinestorepresentselectedfeaturesoftheenvironmentinageneralizedfashion,buttheobjectssuchasindividualpolygons,linesandpointswithinanimagearetypicallyunknown. TheintegrationofremotesensinginformationintoaGISoccursnaturallyinarasterGISbecausebothdatastructuresareapproximatelythesame.Integrationintoavectorsystemrequiressomewhatmoreeffort,butithasbeenrecentlyachievedbyseveralGISandremotesensingvendors.Whenoneisstoringalargeamountofimagedatarelatedtosomeapplicationsintoaspatialdatabasesystem,itisnotpossibletoprocessthedataandtoobtainpolygonalrepresentationsofallobjectsthatwillbeofinteresttousersinadvance.Ontheotherhand,itmightbeofgreatvalueinsuchasystemtohaveavectorrepresentationofcertainimportantobjects.Forthisapplication,oneneedstodesignasystemwithbothrasterandvectorcapabilities,aswellastoolstointerrelatethem.Especially,asinFig.1,weneedtoprocessbothrasterandvectorsimultaneously. Asdescribedabove,whileaspatialinformationsystemmayhaveseveraldatastructures,thesystemcouldbeextendedinseveralwayswhichcorrespondtothefollowingdifferentproperties(seeFig.2): 1)Thestuructureinforma
简介:Airborne3DimagewhichintegratesGPS,attitudemeasurementunit(AMU),scanninglaserrangefinder(SLR)andspectralscannerhasbeendevelopedsuccessfully.ThespectralscannerandSLRusethesameopticalsystemwhichensureslaserpointtomatchpixelseamlessly.Thedistinctiveadvantageof3Dimageisthatitcanproducegeo_referencedimagesandDSM(digitalsurfacemodels)imageswithoutanygroundcontrolpoints(GCPs).ItisnolongernecessarytosurveyGCPsandwithsomesoftwaresthedatacanbeprocessedandproducedigitalsurfacemodels(DSM)andgeo_referencedimagesinquasi_real_time,therefore,theefficiencyof3Dimageis10~100timeshigherthanthatoftraditionalapproaches.Theprocessingprocedureinvolvesdecomposingandcheckingtherawdata,processingGPSdata,calculatingthepositionsoflasersamplepoints,producinggeo_referencedimage,producingDSMandmosaicingstrips. Theprincipleof3Dimageisfirstintroducedinthispaper,andthenwefocusonthefastprocessingtechniqueandalgorithm.Theflighttestsandprocessedresultsshowthattheprocessingtechniqueisfeasibleandcanmeettherequirementofquasi_real_timeapplications.
简介:Thispaperinvestigatesthemathematicfeaturesofnon-linearmodelsanddiscussestheprocessingwayofnon-linearfactorswhichcontributestothenon-linearityofanonlinearmodel.Onthebasisoftheerrordefinition,thispaperputsforwardanewadjustmentcriterion,SGPE.Last,thispaperinvestigatesthesolutionofanon-linearregressionmodelinthenon-linearmodelspaceandmakesthecomparisonbetweentheestimatedvaluesinnon-linearmodelspaceandthoseinlinearmodelspace.
简介:Theprocessingofnonlineardatawasoneofhottopicsinsurveyingandmappingfieldinrecentyears.Asaresult,manylinearmethodsandnonlinearmethodshavebeendeveloped.Butthemethodsforprocessinggeneralizednonlinearsurveyingandmappingdata,especiallyfordifferentdatatypesandincludingunknownparameterswithrandomornonrandom,areseldomnoticed.Anewalgorithmmodelispresentedinthispaperforprocessingnonlineardynamicmultiple-periodandmultiple-accuracydataderivedfromdeformationmonitoringnetwork.
简介:Thefeasibilityofmoni-toringthechangeofcitysettlementusingGPSsurveyinginsteadoflevel-ingisstudied.AfiducialnetworkandamonitoringnetworkareestablishedinNingbocity.TwoperiodsofGPSobservationarecompleted.Somemeasuresaretakenduringtheobser-vationinordertoensuretoobtainthehigh-preciseheightcomponent.TheSaastamoinenmodelisadoptedinthedataprocessingofthedrycomponentpartofthetroposphericdelay.Thewetcomponentchangeofthetropo-sphericdelayisestimatedbystochasticprocessesmodel.WhenBernesesoft-wareisusedtoprocessthedata,themillimeterlevelprecisionofheightmeasuringisachieved.