Advances in Elastic Graph Matching for Frontal Face Verification

Elastic graph matching is one of the most well known techniques for frontal face recognition/verification and one of the few techniques that can be combined successfully with fully automatic face localization and alignment methods. In this paper we propose a series of techniques that enhance the performance of elastic graph matching in frontal face verification by exploiting the individuality of human facial features in many ways. First the use of discriminant analysis in the feature vectors of the graph nodes is explored. The use of the node deformation for discrimination is also proposed. Moreover, the local similarity values at the nodes of the elastic graph, are weighted by coefficients that are also derived from some discriminant analysis in order to form a total similarity measure between faces. We present an algorithm that combines all the above discriminant steps. Moreover, we propose an algorithm for finding the most discriminant landmarks upon a person´s face and a person-specific graph is placed in the spatial coordinates that correspond to these discriminant features. We illustrate the improvements in performance by the proposed advances in frontal face verification using the XM2VTS database