Discriminant Improved Local Tangent Space Alignment Feature Fusion for Face Recognition

Improved Local Tangent Space Alignment (ILTSA) is a recent manifold learning method. In this paper, based on linearization and discriminant extension of ILTSA, a novel feature extraction method named Discriminant ILTSA (DILTSA) is proposed with its theory and algorithm analysis. Based on maximum neighborhood margin criterion and ILTSA, DILTSA can preserve both local within-class and between-class geometry structures. In face recognition application, an augmented Gabor-like complex wavelet transform is proposed, which can efficiently alleviate the illumination and expression variation effect. An approach for face recognition based on the fusion of local and holistic features is developed. Experimental results on Yale and PIE face databases demonstrate the effectiveness of the proposed face recognition method.