資源描述:
《基于稀疏表示的人臉識(shí)別算法研究》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。
1、基于稀疏表示的人臉識(shí)別算法研究江雨大學(xué)物聯(lián)叫工程學(xué)院,江蘇無錫214122摘要:生物將征識(shí)別是通過生物傳感器、計(jì)算機(jī)以及生物統(tǒng)計(jì)學(xué)原現(xiàn)等,依裾人所固冇的生理或行為特征對(duì)個(gè)人身份進(jìn)行鑒定。而人臉識(shí)別是生物特征識(shí)別技術(shù)中的一個(gè)重耍分支,在人機(jī)交互。安全認(rèn)證等領(lǐng)域有著廣泛的應(yīng)川。人臉特征具舍唯-?性,但足人臉圖像也容易受到各種千擾因索的影響,由此對(duì)人臉識(shí)別造成一定的影響。在現(xiàn)打的人臉識(shí)別算法屮,大多數(shù)盂要進(jìn)行閣像預(yù)處理及復(fù)雜的特征捉収,選擇何種特征對(duì)識(shí)別率影響非常大,并且對(duì)遮捫、噪盧等情況缺少魯棒性,這些問題往往使得現(xiàn)行的識(shí)別方法在應(yīng)用屮受到制約
2、。主成分分析法是人臉識(shí)別領(lǐng)域的一個(gè)傳統(tǒng)人法。而?于稀疏表示的人臉識(shí)別是一個(gè)新興K冇效的方法,此力*法是通過稀疏表示來實(shí)現(xiàn)對(duì)人臉的識(shí)別,其有較商的識(shí)別率和較強(qiáng)的魯棒性。本文通過對(duì)Yale-B人臉數(shù)據(jù)庫進(jìn)行識(shí)別測試,可以表明稀疏衣示能夠侖效的提高人臉識(shí)別效米,使識(shí)別率ft大93%。通過實(shí)驗(yàn)可以證明基于稀疏汲示的人臉識(shí)別其宥可行性,同時(shí)能正確識(shí)別人臉圖像并提高識(shí)別率。關(guān)鍵詞:人臉識(shí)別;主成分分析;稀疏表示;魯棒性;識(shí)別率AlgorithmforfacerecognitionbasedonsparserepresentationCollegeofI
3、nternetofThings,JiangnanUniversity,Wuxi.214122Abstract:Thcbiologicalfeaturerecognitionisthroughthebiologicalsensor,computerandbiostatistics,accordingtopersonalidentityidentificationofphysiologicalorbehavioralcharacteristicsofpeopleinherent.Facerecognitionisanimportantbranc
4、hofbiometricidentificationtechnology,inhuman-computerinteraction.Securityauthenticationandotherfieldshaveawiderangeofapplications.Facialfeaturesisunique,butthefaceimagesarealsovulnerabletovariousinterferencefactors,whichaffectthefacerecognition.Intheexistingfacerecognition
5、algorithm,mostoftheneedforimageprocessingandcomplexfeatureextraction,influencethechoiceofthecharacteristicsoftherecognitionrateisveryhighlandthelackofrobustnesstoocclusion,noiseandsoon,thcscproblemsarcoftenmakestheexistingrecognitionmethodsarerestrictedinapplication.Princi
6、palcomponentsanalysismethodisatraditionalmethodforfacerecognition.Thefacerecognitionbasedonsparserepresentationisanewandeffectivemethod,thismethodisthroughthesparserepresentationforfacerecognition,hashighrecognitionrateandrobustness.TherecognitiontestsperformedonYale-Bface
7、database,mayindicatethatthesparserepresentationcanimprovetherecognitionrateeffectively,andmaketherecognitionrateashighas93%.Thcexperimentalresultsshowthatthefacerecognitionbasedonsparserepresentationisfeasible,andcancorrectfacerecognitionandimprovetherecognitionrate.Keywor
8、ds:facerecognition;principalcomponentanalysis;sparserepresentation;robustness;recognition