A two-stage face recognition method is proposed in this paper. In the first stage, we present a hybrid method of GA and gradient-based algorithm for training NMF. The set of candidates is narrowed down with the Euclidean distance in NMF subspace serving as the global similarity. In the second stage, face image is segmented into facial bands, and synergetic approach is employed for further recognition. The similarity between a given region of the query image and a stored sample is obtained as the order parameter which serves as an element of the order vector. A modified definition of the potential function is given, and the dynamic model of recognition is derived from it. The effectiveness of the proposed method is experimentally confirmed.