View-Invariant Action Recognition from Point Triplets | IEEE Journals & Magazine | IEEE Xplore