Kinect-Based Face Recognition and Its Application in MOOCs Production

In recent years, the approaches of human computer interaction (HCI) are in rapid improvement. Face recognition is increasingly remarkable in the field of HCI. Besides gesture, face and facial expressions contains a wealth of information which can help computer understand human´s idea and emotion. Face is an important addition to the language of communication and plays an important role in the exchange between people. This paper focus on a face tracking method based on Kinect and its application in MOOC production. Kinect has the advantage over ordinary camera because it has 2 sensor, an ordinary and a depth sensor. In this paper, a method based on depth information is used for optimizing the face recognition performance in MOOCs recording systems. Taking advantage of the depth information from Kinect, this paper define nose as the first decision point by which system can find face rapidly. Then combine with three significant points (eyes and nose) and Zernike Moments method, we propose a face direction recognition algorithm. At last, we design a MOOCs recording system combining with hand gesture and face recognition, which can switch live stream automatically and make teacher easy to operate the PPT screen cast system.