基于復(fù)合神經(jīng)網(wǎng)絡(luò)的掌紋識(shí)別方法的研究.pdf

1、學(xué)校代碼；三旦星業(yè)分類號(hào)：壁!盟研究生學(xué)號(hào)：密級(jí)：⑨東北嶧葒大季碩士學(xué)位論文200421460基于復(fù)合神經(jīng)網(wǎng)絡(luò)的掌紋識(shí)別方法的研究ResearchofPalmprintIdentificationMethodusingCompoundNeuralNetwork作者：王淑華指導(dǎo)教師：呂英華教授孔俊副教授學(xué)科專業(yè)：計(jì)算機(jī)軟件與理論研究方向：數(shù)字圖像處理學(xué)位類型：學(xué)歷碩士東北師范大學(xué)學(xué)位評(píng)定委員會(huì)2007年5月AbstractToday喇th

2、therapiddevelopmentofinformationandInternetallovertheworld，informationsecurityespeciallypersonalidentificationtechnologybecomesextraordinarilyimportantandmusthavewidelyappliedperspectiveHereinto，becauseofitsuniquefeature

3、sofstabilityuniquenessandconveniencethebiologicalfcanlr弓identificationtechnologyhasbeenappliedmoreandmorebroadlyAsakindofveryimportantbiologicalfcalureidentificationtechnologypalmprintautomaticidentificationwillhaveconsi

4、derablepotentialprofitingfromsuchfbanlIeofpalmprintaseasyaoqujsiti∞，majorfeaturebeingapparentstabilitydivisibilityandSOonandfromcertainadvantagesofpalmprintautomaticidentificationsystemsuchasdirectness，kindnessconvenienc

5、e，niceuniquenessandwidelyappliedscopeHowever，duetothecomplexityanddiversityofpahnprinttheresearchandapplicationofbiologicalfeatureidentificationbasedonpalmprintstillstayondevelopingthealgorithmoffeatureabstractinganddesi

6、gningclassifierandtheexistingpalmprintidentificationtechnologymainlybasedonthetexturestructuralorgeometryfeatureofpalmprintHavingthorougidyresearchedtheexistingpalmprintidentificationtechnologyinthispaperwegiveanovelalgo

7、rithmofkeypointslocationandproposeahierarchicalmultifeatureschemetofacilitatecoarsetofinematchingforefficientandeffectivepalmprintrecognitionInourapproachtwolevelsoffeaturearcdefined：geometryfeaturebasedOndistance(1evel1

8、feature)andtexturefeaturebasedonZemikemoment(1evel2feature)Fordifferentfeatures，WCadopttwokindsofdifferentneuralnetworkandthencombinethemintoonerecognitionsystemeffectivelyFinallytheexperimentalresIlltdemonstratesthefeas

9、ibilityandefficiencyoftheproposedsystemBaseonthethoroughresearchandcomparisonoftheexistingpahnprintrecognitiontechnologythemaincontributionsofthispaperconclude：(1)GiveanovelalgorithmofpalmprintkeypointslocationandROIextr

10、aon(2)UsetwoalgorithmsofgeometryfeatureabstractingbasedondistanceandtexturefeatureabstractingbasedonZernikemomenttoabstractthetwolevelfeaturedefinedabove0)Proposeanovelmultifeaturepalmprintidentificationtechnologybasedon