Locating facial features by robust active shape model

Active shape model statistically represents a shape by a set of well-defined landmark points and can model object variations using principal component analysis. However, the shape generated by standard active shape model is unsmooth when the test sample has a large variation compared with the training images. In this paper, we introduce a robust active shape model for facial feature location. First, a color information and 2-dimension based local feature model is presented to characterize salient facial features, such as the eyes and the mouth. Then, a regularized principal component analysis based shape model is proposed to construct a smooth global shape. We evaluate our approach on a challenging dataset containing 2,000 well-labeled facial images with a large range of variations in pose, lighting and expression. Experimental results demonstrate the efficiency and effectiveness of the proposed approach.