基于人臉信息識(shí)別的信息安全技術(shù)研究.pdf

1、東北林業(yè)大學(xué)碩士學(xué)位論文基于人臉信息識(shí)別的信息安全技術(shù)研究姓名：陳愛軍申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：機(jī)械設(shè)計(jì)及理論指導(dǎo)教師：王述洋2002.5.1ABSTRACTFacerecognitionisachallengesubjectinthefieldofpatternrecognitionandithasimportanttheoryandappliancevalueInthispaperwefirstintroducethecurrentr

2、esearch，methodsandtrendonfacerecognitionThenwerespectivelydiscussfaceimagepreprocessing，facefeatureextractionandthedesignofclassifiersInthecourseofimagepreprocessing，digitalfaceimagesarecapturedwithanumeralcsaneraandthes

3、eimages8Jetransformedt08bitfrom24bitThenthenoiseinfaceimagesareremovedbyusingthemethodofmedianfilterandthefaceimagesarebinarizedFinallythefaceineachfaceimagearesegmentedwiththemethodofintegrationprojection，thenstandardiz

4、edStandardfaceimagesareformedthrouIghtheabovementionedprocessingDuringthefeatureextraction，forthosestandardfaceimages，regardingthebetweenclassscattermatrixasgeneratingmatrix，weextractthealgebraicfeaturesoffaceimagesthrou

5、ghK—LtransformandsingularvaluedecompositionInthecourseofclassifiersdesign，consideringthatthesingleclassifierhasnothighrecognitionrate。weconstructacombiningclassifierwithaminimumdistanceclassifierandafuzzyNNclassifiertoim

6、provetherecognitionrateFinallytheexperimentresultsoffacerecognitionarepresentedandanalyzedTheresultsindicatethattheextractedfacefeaturesarevalidandthedesignofclassifieriSsoundandefficientKeywords：facerecognition，algebraf