Based On Improved Principal Component Analysis For Face Recogniiton Algoirthm Research

Face recognition is an active and challenging topic in patternrecognition fields, in which extraction of the effective discrimiant features isa very important step. In the past decades, a lot of feature extraction methods,however, principal component analysis(PCA), which is developed in thispaper, is the basic method in face recognition. All the methods proposed inthis dissertation are proved to be effective and feasible in face recognition.The principal research work and novelties are listed as follows：1. Some main problems corresponding to face recognition are propose in thefirst part of this paper, for example, background study, state of the field, andresearch method, and so forth.2. The PCA theory is discussed in the second part of this paper, and thecharacteristic of the PCA, the main shortcoming is analyzed.3. A face recognition method based PCA feature reconstruction is proposedafter a completely research. Some experiments prove that the new algorithmis effective.4. The PCA is combined to the maximum scatter difference discriminantanalysis method to improve the face recognition performance, and the newface recognition method based on the second reduction dimension of PCAfeature is proposed, whose feasibility and effectiveness is demonstrated bythe experiment results on several famous face databases.