Frontal human detection based on body structure measured by using depth image

In this paper, we propose frontal human detection method by using real-measurement of object by using depth camera. Two geometrical features of human body are used to detect the human orientation, such as shoulder width and depth difference between head and chest. The validity of proposed algorithm is confirmed by using real sequences. The proposed algorithm would be very useful in early phase of gesture recognition system to decide the human to interact with HCI system.