An Improved Active Shape Model for Facial Feature Location

Abstract

Active shape model (ASM) has been widely used to extract the facial features of a face image. In this paper, we propose an improved ASM algorithm for face image features extraction. The conventional ASM uses a single face model to represent a face, so it may not be able to capture the local shape variations effectively under different facial expressions. In our proposed approach, we use two individual feature models: one includes the face contour and mouth; the other one includes the two eyebrows, two eyes and nose. Firstly, we get the shape model and the gray-level model for the two individual features separately; then, we use the above shape model and the gray-level model to interpret the two individual feature models of a novel face image; finally, we combine the two shape models to represent the face. Experimental results show that our proposed method can extract the facial features more accurately than tradition ASM.