Viewpoint Invariant Human Re-Identifi cation in Camera Networks Using Pose Priors and Subject-Discriminative Features