基于人工神經(jīng)網(wǎng)絡(luò)的水面運(yùn)動(dòng)目標(biāo)識(shí)別技術(shù)研究.pdf

1、哈爾濱工程大學(xué)碩士學(xué)位論文基于人工神經(jīng)網(wǎng)絡(luò)的水面運(yùn)動(dòng)目標(biāo)識(shí)別技術(shù)研究姓名：李薇申請(qǐng)學(xué)位級(jí)別：碩士專(zhuān)業(yè)：船舶與海洋結(jié)構(gòu)物設(shè)計(jì)制造指導(dǎo)教師：萬(wàn)磊20070101哈爾濱T稗大學(xué)碩十學(xué)仿論文AbstractIntelligenceandhi曲speedarealwaysimportantdirectionsofshipresearchAsthespeedofshipincreasing，weneedshipmoreintelligentVisi

2、onisanimportantapproachtoacquireinformationTherefore，thedevelopmentofvisionintelligenceisimportanttoenhanceship’Sindependentabilitywhichmakesshiprecognizetargetexactlyresponsetocorrespondinginformationandcompletethetasks

3、uccessfullyFirstlyconsideringthecharacteristicofmovingobjectsabovethewaterwedosimulativeexperimentsintheflumeImagesofshipmodelsindifferentcircumstancesandattitudesareacquiredtobuildanimagedatabaseSecondly，aftermedianfilt

4、eringtheimages，onthebasisofthereal—timecharacteristicofmovingobjects，afeasibleandeffectivemethod一OtsuispresentedtosegmentimagesFinallyweacquiredsatisfactoryeffectofsegmentationTheeigenvectorsofimagesalegotthroughthefeatu

5、reextractionandfeatureselection，whicharethesixfeaturevectorsThrough300flamesoffourobjects’images，weknowthatthesixfeaturevectorshadbetterclusteringeffecttocongenerobjectsandseparableeffecttodifferentkindsofobjectsFinallyw

6、emainlydoresearchonartificialneuralnetworkforidentificationConsideringthelimitationoftraditionalBPneuralnetworkweuseanewParticleSwarmOptimizermethodAfteranalyzingthealgorithmictheoryandparametersetting，wedesignedtheprogr

7、amExperimentalresultsshowthatParticleSwarmOptimizerhasagoodastringencyandcanavoidthedefectofBPneuralnetworkColligatingParticleSwarmOptimizerandBPneuralnetwork，wecanacquiresatisfactoryeffectAccordingtothecomplexofthevisio

8、nintelligenceandinformationdisposingthepaperisonlythegroundworkHowever∞theavailableexploremethodsmentionedinthepaperlaythefoundationforthefurtherworkKeyWords：Movingobjectabovewater，Otsuautomaticsegmentation，BPneuralnetwo