Linear Subspace Recognition Based On Gabor Wavelet Decomposition

Face recognition is an active area of research field by far, which has unique advantages and widely paid attention in research field as a biometric identification technology. Because of the particularity of human face, resulting as several problems among the theoretical and practical recognition applications, it should be further improved and perfected.This paper mainly studies the face recognition algorithms and combines some existed face recognition algorithms. A face recognition system includes face detection, feature extraction and classification recognition ect, and feature extraction is the most significant part of face recognition system and determines the final recognition rate. However, the Gabor wavelet has outstanding advantages in image feature extraction, using the Gabor wavelet as face image feature extraction, combines with modular PCA and LDA linear discriminant analysis method as the face recognition method in this paper.Firstly, according to the feature extraction based on Gabor wavelet method, using the Gabor wavelets function with 5 scales and 8 orientations constructed by Gabor wavelet applied convolution with face image, effective access to the face feature information with the different scales and orientations. In order to reduce the feature dimension, choose one of the scales and orientations of Gabor facial features as the following recognition feature.Secondly, it highlights the widely used principle component analysis (PCA) and linear discriminant analysis (LDA) algorithms, based on the face recognition of linear subspace using modular PCA method makes dimension reduction to Gabor facial features, then use linear discriminant analysis LDA feature classification, and finally use the nearest neighbor classifier to match facial feature recognition, which not only improves the recognition rate also makes it easier to classify facial feature recognition.Finally, this paper adopts the international standard ORL database as human face samples are given recognition rate by the experimental results of the proposed method and compared with other linear subspace face recognition methods. Experiments show that the face recognition method based on the linear subspace with Gabor wavelet in this paper a better recognition results.