基于人工神經(jīng)網(wǎng)絡(luò)技術(shù)的高光譜遙感淺海水深反演研究.pdf

1、分類號UDC密級研究生學(xué)分論文基于人工神經(jīng)網(wǎng)絡(luò)技術(shù)的高光譜遙感淺海水深反演研究研究生姓名旗蕓掘指導(dǎo)教師姓名毖皇握副熬授申請學(xué)位級別亟專業(yè)名稱物堡連注堂(鎏注撞壟2論文答辯日期2QQ耋生壘且!蘭目學(xué)位授予日期2QQ量生且中國海洋大學(xué)量主△三!!絲塑墊堇查塑壹堂堂墨壁些塑查鱉墾塑塑塞Studyofthehyperspectralremotesensingofshallowwatersbathymetrywithartificialneura

2、lnetworktechnologyAbstractItisimportanttodevelopthefastandaccuratemethodsofbathymetryforeconomicdevelopment，militaryaffairsandnationaldefeneeAmethodofhyperspectralremotesensingofshallowwaterbathymetrybasedonartificialneu

3、ralnetwork(ANN、techniqueisstudiedinthisthesisBasedontheinherentopticalpropertiesoftherelevantwaters，hyperspectralremotesensingreflectctance(R目simulationiscarriedoutbysend—analyticalnumericalsimulationtechniqueoftheradiat

4、ivetransferequation0nethree1ayerartificialneuralnetworkisdevetopedtoderivedepthofshallowwatersbasedonthesimulatedhyperspectralremotesensingreflectctancedataAmethodfordistinguishingopticalshallowwatersandopticaldeepwaters

5、isproposedbasedtheinherentopticalpropertiesofthewatersandthebottomfeaturestoguaranteethevalidityofthesimulateddataIntheprocessofmodelingANNinordertoacceleratetheback—propagationlearningprocesstheinputsoftheANNareprocesse

6、dinadvancebyprincipalcomponentanalysisComparedwiththeresultoftheoptimizationanalysismodeltheANNshowsgoodperformanceingeneralizationBesides，apreliminarystudyofthebottomclassifyingofshallowwaterbathymetrybyspectralderivati