Incorporating corner information for mouth feature extraction

Mouth feature extraction is of crucial importance for face recognition. Deformable template is an efficient model for this task. However, it suffers from the problems of local minima and low convergence speed. To fill this gap, in this paper, a new energy function is proposed, which emphasizes on corner information ignored in previous methods. Thus, the energy function can reflect the properties of mouth feature effectively. Experimental results show the new method is robust.