Target Image Extraction for Face Recognition Using the Sub-Space Classification Method

This paper proposes a scheme that offers robust txaaction of the target face image in standard view, which is defined using internal facial features as steady reference points. The scheme is characterized by two steps: facial feature detection using color image segmentation, and target image selection from among the candidates usil~g the subspace classification method. The scheme's flexibility has been confirmed in experiments under a wide range of image acquiat~on conditions.