Human action recognition using a depth sequence key-frames based on discriminative collaborative representation classifier for healthcare analytics

Using deep map sequence to recognize human action is an important research field in computer vision. The traditional map-based methods have a lot of redundant information. Therefore, this paper proposes new feature expression method based on discriminative collaborative representation classifier, which highlights the time features. In paper, energy established according shape and characteristics body obtain information body. Then projected onto three orthogonal axes spatialtemporal map. Meanwhile, order solve problem high misclassification probability similar samples by classifier (CRC), CRC (DCRC) proposed. takes into account influence all training each kind coefficient, it obtains highly improves discriminability samples. Experimental results MSR Action3D data set show that redundancy key-frame algorithm reduced, operation efficiency improved 20%-30%. proposed reduces extraction rate It not only preserves spatial through field, but also records temporal complete way. What?s more, still maintains recognition accuracy with