Local Block Representation for Face Recognition

Face expression analysis and recognition play an important role in human face emotion perception and social interaction and have therefore attracted much attention in recent years. Semi-Supervised manifold learning has been successfully applied to facial expression recognition by modeling different expressions as a smooth manifold embedded in a high dimensional space. However, the best classification accuracy does not necessarily guarantee as the assumption of double manifold is still arguable. In this paper, we study a family of semi-supervised learning algorithms for aligning different data sets that are characterzied by the same underlying manifold. The generalized framework for modeling and recognizing facial expressions on multiple manifolds is presented. First, we introduce an assumption of one expression one manifold for facial expression recognition. Second, we propose a feasible algorithm for multiple manifold based facial expression recognition. Extensive experiments show the effectiveness of the proposed approach.