資源描述：
《兩種圖像鑒別特征提取算法研究》由會(huì)員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫(kù)。
1、摘要生物特征識(shí)別技術(shù)，是當(dāng)前生物測(cè)定學(xué)領(lǐng)域中最具代表性和最富有挑戰(zhàn)性的重要研究?jī)?nèi)容。生物特征識(shí)別的研究如今進(jìn)入一個(gè)新的發(fā)展高峰，各種新的技術(shù)和方法被持續(xù)不斷地開(kāi)發(fā)出來(lái)。其中圖像鑒別特征提取算法備受研究人員關(guān)注。如何提取有效的圖像鑒別特征是生物特征識(shí)別技術(shù)研究的熱點(diǎn)。本文提出兩種圖像鑒別特征提取算法研究，主要工作如下：(1)對(duì)圖像鑒別特征提取算法進(jìn)行了概括總結(jié)。其中對(duì)非線性鑒別特征提取算法和頻域分析下的鑒別特征提取算法進(jìn)行深入分析，重點(diǎn)研究了這兩種鑒別特征提取算法的性質(zhì)。(2)重點(diǎn)研究頻域分析下的非采樣Cont
2、ourlet小波變換。非采樣Contourlet小波變換具有多尺度，時(shí)移不變性以及多方向性等優(yōu)點(diǎn)，對(duì)圖像的邊緣和輪廓有很好的逼近表示。將非采樣Contourlet小波變換和非線性鑒別特征提取技術(shù)相結(jié)合，提出基于非采樣Contourlet小波變換的非線性鑒別分析，即基于非采樣Contourlet小波的掌紋圖像非線性鑒別特征提取算法，該算法有效的提取圖像鑒別特征。(3)分析研究了張量鑒別分析，在掌紋數(shù)據(jù)庫(kù)、AR人臉圖像數(shù)據(jù)庫(kù)和FERET人臉圖像數(shù)據(jù)庫(kù)上進(jìn)行了張量鑒別特征提取算法實(shí)驗(yàn)。關(guān)鍵詞:生物特征識(shí)別；圖像鑒別
3、特征；非線性鑒別特征提?。环遣蓸覥ontourlet小波變換；張量鑒別分析IABSTRACTBiometicRecognitionTechniqueisthemostrepresentativeandchallengingresearchinthecurrentbiometricfields.Biometricsresearcheshavereachedanewdevelopmentpeakinwhichallkindsofnewtechnologyandmethodsaredeveloped.Nowaday
4、s,imagediscriminantfeatureextractionalgorithmhasbeenfocusedonbymanyresearchers.Howtoextracttheimagediscriminantfeatureeffectivelybecomesaresearchhotspot.Twokindsofimagediscriminantfeatureextractionmethodsareresearchedinthispaper.Morespecifically,ourcontribu
5、tionsareasfollows:(1)Theimagediscriminantfeatureextractionmethodsaresummarized.Thenonlineardiscriminantfeatureextractionalgorithmsandthediscriminantfeatureextractionalgorithmsbasedonthefrequencydomainanalysisarestudiedexplicitly.Especiallythenatureofthetwok
6、indsofdiscriminantfeatureextractionalgorithmsisanalyzed.Combingthetwotechnologies,theimprovedfeatureextractionalgorithmsareproposed.(2)Theapplicationofthenon-subsampledcontourlettransform(NSCT)isstudiedinthispaper.Thenonsubsampledcontourlettransformhaschara
7、cteristicsofgoodmultiresolution,shift-invarianceandhighdirectionality.Itcangiveanasymptoticoptimalrepresentationofedgesandcontoursinimage.Anonlineardiscriminantanalysisbasedonthenon-subsampledcontourlettransform(NSCT)forpalmrecognitionisproposed.(3)Thetenso
8、rdiscriminantanalysisisalsoanalyzedandstudiedinthispaper.Inthispaper,thetensordiscriminantanalysiswasconductedonthepalmdatabase、ARfacedatabaseandFERETfacedatabase.Keywords：BiometricRecognition;ImageDis
