Hierarchical Face Recognition Based on SVDD and SVM

Current face recognition methods are mainly based on face database. Face recognition task in natural environment demands face recognition algorithm has the rejection capability for the face samples out of face database, but existing methods lack this rejection ability for non-target samples. In this paper, a new hierarchical face recognition algorithm is proposed which can reject non-database test samples and classify model samples within database exactly. One-class recognition characteristics of support vector data description is firstly utilized to rejection recognition and then the excellent classification property of support vector machine is employed to recognize the accepted face samples. By way of simulated experiments, the effectiveness of proposed method is verified.