Discriminant analysis of haar features for accurate eye detection

The efficient and discriminating feature extraction is a significant problem in pattern recognition and computer vision. This paper presents a novel Discriminating Haar (D-Haar) features for eye detection. The D-Haar feature extraction starts with a Principal Component Analysis (PCA) followed by a whitening transformation on the Haar feature space. A discriminant analysis is then performed on the reduced feature space. A set of basis vectors, based on the novel definition of the within-class and between-class scatter vectors and a new criterion vector, is defined through this analysis. The D-Haar features are derived in the subspace spanned by these basis vectors. We then present an accurate eye detection approach using the D-Haar features. Experiments on Face Recognition Grand Challenge (FRGC) show the promising discriminating power of D-Haar features and the improved detection performance over existing methods.