Hybrid N-feature extraction with fuzzy integral in human face recognition

J. Haddadnia1, K. Faez1
1Electrical Engineering Department, Amirkabir University of Technology슠, Tehran, Iran

Tóm tắt

This paper introduces an efficient method for human face recognition that employs a set of different kinds of feature domains with RBF neural network classifiers, and which is denoted the hybrid N-feature (HNF) human face recognition. A combination of RBF neural network classifiers with fuzzy integral has been proposed to achieve face classification with higher performance. The feature extractor projects the face images in each appropriately selected transform domain in parallel. Experimental results on the ORL database confirm that the proposed method lends itself to higher classification accuracy relative to existing techniques.

Từ khóa

#Humans #Face recognition #Neural networks #Fuzzy neural networks #Feature extraction #Data mining #Face detection #Fuzzy systems #Principal component analysis #Discrete cosine transforms

Tài liệu tham khảo

10.1121/1.401205 10.1016/0888-613X(95)00132-Z 10.1109/AFGR.2002.1004175 10.1109/ICIP.1996.560536 10.1109/36.752193 10.1109/ICIP.2001.959221 10.1109/ANZIIS.1996.573930 10.1109/34.667881 haddadnia, 2001, Human Face Recognition with Moments Invariant, IEEE-EURASIP Workshop on Nonlinear Signal and Image Processing 10.1109/ICPR.2000.906118 daugman, 2001, Face Detection: A Survey, Computer Vision and Image Understanding, 83, 236, 10.1006/cviu.2001.0921 10.1016/S0031-3203(99)00104-1 10.1162/jocn.1991.3.1.71