Person authentication using 3D human motion

This paper presents a novel approach to identify and/or verify persons by using three-dimensional dynamic and structural features extracted from human motion depicted on image streams. These features are extracted from body landmarks which are detected and tracked when the person is asked to perform specific movements, representing the dynamics of specific parts of the body, as well as the structural traits formed by the pose of the person. Gaussian mixture model (GMM) based systems are tested on a dataset containing arm movements. Experimental results confirmed that the proposed approach is promising for person authentication tasks.