Research Of Face Recognition Based On 2D-PCA

With the rapid development of information technology, information security becomesmore and more important today. Some application domains need efficient automatic personalidentification technology, such as Electron Commerce, Electron Bank, Network Security andso on. Biometrics gets more and more widely application in view of its stability, uniquenessand convenience. From the warranty of entrance to lock criminal in crowd, all are related tothe market of application and trend of future of this technology.Biometrics combines the information technology with biology technology, which useshuman inherent biological characteristics such as face, palm-print, and iris, behavioral characteristicssuch as gait, signature and speech to confirm personal identity for replacing orstrengthening the traditional personal identification approaches. Face recognition has verylarge academic and practical values. In daily life, people knowing each other uses at most ofperson's face. The visual information reflected by face has important meaning and impactbetween people's inter communion and intercourse. Because of its extensive and appliedrealm, face recognition technique has got the extensive concern with study in near three decadesand become the most potential method of identity recognition.This paper proposed one kind of Improved two-dimensional principal componentsanalysis which I2D-PCA ,the personal face recognition methods, on having done the basisstudying with thorough ingredient analysis to the principal component .firstly, to carries onthe pretreatment to the face image, the use of the training sample collection, the I2D-PCAmethod which presents in this paper calculate the principal component, the definition characteristicspace, a training sample of each characteristic of the coefficient matrix; secondly, theresearch got a group of characteristic coefficient matrix of detecting that on the samplewith Carries on the similar pretreatment operation in the verification stage to the examinationsample , and maps it above the characteristic space which calculates based on the trainingsample comes; In the end, the experiment were using the minimum from classificationimplement to be in progress for identification calculating distance between the moduluschecking and training the sample book characteristic .We respectively in ORL database Yale B database, the AMP database has carried out anexperiment verified 2D-PCA face recognition methods validity which this paper proposed, theexperiment had indicated that the method of this paper proposed has the good recognition effectin the calibration database with carrying on the contrast between the method which proposedto this paper and traditional 2D-PCA.