数字近景摄影测量中人工靶标检测与识别技术研究-计算机科学与技术专业论文.docx

数字近景摄影测量中人工靶标检测与识别技术研究11万方数据AbstractAbstractIndigitalclose-rangephotogrammetry,speciallymadeartificialtargetscanhelpalottoquicklyandeasilycompletethephotogrammetricmissionslikecorrespondingpointmatching，3Dreconstruction，camerapositionandorientationetc．However,duetotheincreasementofengineeringapplicationsincomplexenvironmentandregionslackoftexture，the shortcomings ofartificialtargetsinexisting measurementsystemhasseriouslyrestrictedreflectingitssuperiorities．Therefore，ithasbecomeanurgentneedtoensurethemeasurementaccuracyandtheworkefficiencytodesignakindofartificialtarget of simple characters，convenientforlayout andeasyforextractionalgorithm；effectivelyreducethetimecomplexityandimprovetherobustnessofthealgorithmoftargetdetectionandrecognition，andenhancetheabilityofdataprocessing．Hence，thisthesisproposeanovelartificialcodedtargetandacomprehensivealgorithmofartificialtargetdetectionandrecognitionbased ontemplatematching．Theresearchworkcanprovidefineffectiveimprovementschemefordigitalclose．rangephotogrammetryandhasimportanttheoreticalsignificanceandapplicationvalue．Firstly,anovelartificialcodedtargetisproposed．Differentfromthecommoncircle．basedtarget，theproposedtargettotallyseparatethetargetboxandcodingbox，whichcannotontyguaranteesufficientcodingnumber,butalsoavoidtheinterf．erencetothelocationofthefeaturepoints．Meantime，withthesimpleanduniquegeometricproperties，itiseffortlesstodistinguishbetweenthetargetsandnaturalbackground．Furthermore，therobustnessofthesubsequentalgorithmoftargetdetectionandrecognitioncanbeenhancedsignificantlybyusingtheverificationcodetechnologyinthedecodingprocess．Secondly,acomprehensivealgorithmofartificialtargetdetectionandrecognitionbasedontemplatematchingisproposed．Thetraditionalmethodswhichisbasedonellipse fiuinghavemanydisadvantagesincludinglargeamountofcomputation，seriousdependenceonedgefeaturesandeasytobedisturbedbyimagenoise·Consideringtheseproblems，thisthesisHsestheshape-basedtemplatematchinglTT万方数据数字近景摄影测量中人工靶标检测与识别技术研究algorithmasthecore，c