Abstract—This paper presents a method using a compound classifier which exploits both Gabor features and Fourier features extracted from input images. Gabor wavelet is used to capture facial relevance characters, i.e. Gabor features, in different scales and directions. Given the high-dimension of Gabor features, we employ PCA method to refine them. Then, Fourier features are exploited by applying the Discrete Fourier transform to images and then retaining the low-frequency coefficients, which contain the facial contour information. Thus, 2 classifiers are obtained: one based on Gabor feature, the other based on Fourier feature. Finally, the 2 classifiers will be integrated together to form a final classifier. We evaluate this method using ORL face database. Experimental results illustrate the performance of this method compared with PCA and FLDA.
Index Terms—2-layer classifier, gabor feature, fourier features, PCA.
The author is with the Computer Center, East China Normal University (e-mail: yxlyxl007@qq.com, Tel.: 086-18801966730).
Cite: Xiaolu Yang and Junyi Zhao, "2-Layer Classifier for Facial Recognition," International Journal of Future Computer and Communication vol. 1, no. 3, pp. 267-269, 2012.
Copyright © 2008-2025. International Journal of Future Computer and Communication. All rights reserved.
E-mail: editor@ijfcc.org