环境下弱小目标检测与识别技术研究

南京航空航天大学硕士学位论文复杂环境下弱小目标检测与识别技术研究姓名：谭晓宇申请学位级别：硕士专业：模式识别与智能系统指导教师：陈谋20080201iVC++OpenCViiABSTRACTThedetectionandrecognitionofdimandsmalltargetisahotresearchsubjectinthefieldofimageprocessandtargetdetectionandrecognition.Thisresearchalsohasanimportantstrategicapplicationvalue.Inthisdissertation,detectionandrecognitionofdimandsmalltargetinimagesarestudied.Dimandsmalltargetsaredetectedbyprocessingimagesandberecognizedwithinformationtechnique.Fistofall,thebasiccharacteristicsoftheimageswithsmalltargetsarestudied.Thesecharacteristicsincludethemodelsofimagenoise,spacebackgroundandsmalltargets.Theimagepre-processingbeforetargetsdetectionisstudiedtoo.Then,“DetectBeforeTrack”algorithmisstudied.Itisusedfordetectingdimandsmalltargetsincomplexbackgroundimageswithweeknoise.Asthedimandsmalltargetscannotbedetectedwithasingleimageprocessingtechnique,twonewmethodsofdimandsmalltargetsdetectionareproposed.Oneisbaseonthewavelettransformalgorithmandinformationfusiontechnology.Theotherisbaseonthewavelettransformalgorithmandmathematicalmorphology.Simulationresultsshowthatthetwonewalgorithmscanobtainthegooddetectionresults.Following,“TrackBeforeDetect”algorithmisstudied.Itisusedfordetectingdimandsmalltargetsinsimplebackgroundimageswithstrongnoise.Inallusiontotheflawsofgeneraldynamicprogrammingalgorithmwithmaximizingenergy,anewimprovedmethodisproposed,whichisbasedonthemoststeadystates.Simulationresultsshowthattheproposedalgorithmcanwellsolvetheenergyscatteringproblem.Andthen,recognitionalgorithmofdimandsmalltargetsisstudied.Becauseofthelackofthesmalltargets’information,itisdifficulttorecognizesmalltargetsandtherecognitionreliabilitiesalwaysarelow.Anewimprovedmethodofsmalltargetsrecognitionisproposedwhichisbaseonthetheinformationoftargets’velocity,grayandtracks.Simulationresultsshowthatrecognitionreliabilitiesareraisedobviously.Finally,akindofsoftwareofdetectionandrecognitionofdimandsmalltargetisdevelopedwithVC++andOpenCV.Severalbasicdimandsmalltargetsdetectioniiiandrecognitionalgorithmsandnewmethodsareimplementedbyit.Theeffectofthesenewmethodsinthisdissertationisvalidatedbyusingthesoftware.KeyWords:DimandSmallTarget,TargetDetection,TargetRecognition,WaveletTransform,MathematicalMorphology,DynamicProgramming,InformationFusionvii1.1...................................................................72.1.................................................................................................113.1DBT.........................................................................................143.2.........................................................153.3.............................................................................163.4....................................163.5.....................................................................................173.6.............................................................................................173.7.............................................................................................183.83.6(a)......................................................................183.9.....................................................................193.10...................................................................203.11...............................................203.12...............................................................................213.13...................................................................................223.14...........................................................................................273.15.......................................................................273.16...............................................................................283.17...............................................................................293.18...................................................................................303.19.......................................................................................323.20.......................................................................................323.21...............................................................................323.22...............................................333.23...............................................333.24HVD................................................343.25HVD...........................................................343.26..............343.27..............34viii4.1.............................................................................................374.2.....................................................................394.33×31()kDx−..............................................................................404.4SSDP...........................................................................414.5-...........................................................................424.6.....................................................................................434.75×5...........................................................