Labeled graphs and dynamic link matching for face recognition and scene analysis