基于LOGGABOR滤波的掌纹识别算法研究

《基于LOGGABOR滤波的掌纹识别算法研究》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、西南交通大学硕士学位论文基于LOG-GABOR滤波的掌纹识别算法研究姓名：温长芝申请学位级别：硕士专业：信号与信息处理指导教师：张家树20080501西南交通大学硕士研究生学位论文第1I页I_llI_-I-l_●-I--__●Ill●l_●l●一I—I●IllIllI__●l--I__●__●_●l__lll_●-Il●__I—ABSTRACTWiththerapiddevelopmentofinformationtechnologyandinternetallovertheworld，informationsecurityisext

2、raordinarilyimportant，especiallyinpersonalidentificationtechnology．Biometricstechnologyisgainingmoreandmoreattentionrecently，becausebiologicalfeaturehasuniquestability，uniquenessandconvenience．Asanewbiometrictechnology，palmprintrecognitionhasbeenfocusedonintheseyearsdue

3、toitsadvantagessuchaseasycapturing，abundantimageinformation，high-rateacceptancebyusers，difficultyinfakingandfewerchancestobeinterferedbynoiseetc．．Thethesismainlyinvestigatesinpalmprintrecognitionfeatureextraction．Themaincontentsandoriginalitiesinthisthesisarelistedasfol

4、lows：1、Threealgorithmsaboutlocatingandsegmentingpalmprintimageintoregion—of．interestareanalyzedandtheirperformancesarecomparedincomputersimulations，inwhichthealgorithmbasedonmorphologicoperatorisused．2、AnovelpalmprintrecognitionmethodusingGaborwaveletsand2一Dimensionalpr

5、incipalcomponentanalysis(2·DPCA)andPCAisusedfurtherfeaturedimensionalreduction．SimulationresultsshowthattheproposedmethodissuperiortothatofGaborplusPCAmethodinpalmprintrecognitionandidentification．3、Anoptimal2一DorthogonalLog—Gaborfilterisconstructedbytwo2·DLog—Gaborfilt

6、erswithorthogonalorientations．Thefilteristhenusedtoextractorthogonalphasefeaturesfrompalmprintimages．HammingdistanceiSusedasaclassifierinthismethod．ExperimentalresultsshowedthattheperformanceofthisnovelLog·Gaborfilterismuchsuperiortothoseofalgorithmsbasedon2-DGaborfilte

7、rs．4、Anew2-DmixtureLog-Gaborfilterisconstructedbyanumberofweighted2-DLog—Gaborfilterswithdifferentorientations．Thisnew西南交通大学硕士研究生学位论文第1lI页-_●●_●●●●_--_●●-●-_●___●●-_____________●_-●_-l___-_●__●●_一II●I_●-●●-●●_-●_____●-_●-_●--_-___●l_-_l___-●___一filterisusedtoextractmixt

8、urephasefeaturefrompalmprintimages．Experimentalresultsshowthatcorrectrecognitionratesarefurtherimprovedbyusing