Face recognition using independent gabor wavelet features

Abstract

We introduce in this paper a novel Independent Gabor wavelet Features (IGF) method for face recognition. The IGF method derives first an augmented Gabor feature vector based upon the Gabor wavelet transformation of face images and using different orientation and scale local features. Independent Component Analysis (ICA) operates then on the Gabor feature vector subject to sensitivity analysis for the ICA transformation. Finally, the IGF method applies the Probabilistic Reasoning Model for classification by exploiting the independence properties between the feature components derived by the ICA. The feasibility of the new IGF method has been successfully tested on face recognition using 600 FERET frontal face images corresponding to 200 subjects whose facial expressions and lighting conditions may vary. © Springer-Verlag Berlin Heidelberg 2001.