Journal of Electronic Measurement and Instrument (电子测量与仪器学报) 2013/27:9 PP.817-822
This paper proposes an algorithm which represents the human face from both global and local part. First of all, Gabor wavelet can express facial features from different directions and different scales efficiently, it can highlight the significance of local part. Then, principal component analysis (PCA) extracts the facial contour information which can make up the shortages of Gabor wavelet in extracting the facial global features. Therefore, we use PCA to extract the facial global features and use Gabor wavelet to extract local features whose dimensions are reduced by PCA. Finally, these features are fused as the general features in face recognition. The experiments show that the effect of the feature fused by Gabor and PCA is superior to that of the single feature on the face recognition. In the case of three images of the same person are served as the training samples, the highest recognition rate can reach 96.79%.