Posture Estimation Of Human Head Using Kinect

In recent years, facial recognition and posture estimation of human head become increasingly important in many areas, such as safe driving, authentication and so on. But the traditional method using 2-D color image to recognize the facial features of human head and then tracking the changes of coordinates of these facial features has some problems. For example, it could be easily affected by make-up, environment or lighting.In order to solve the above problem, in this research, we propose the 3-D posture presumption technique of human head. First, we obtain the 3-D point cloud image of human face by a new device named Kinect, and then we use this 3-D point cloud image to presume the posture of human head. In order to presume the posture, of human head, we introduce the principles of the laboratory equipment and specific experimental method named ICP algorithm. The experimental result proved the validity of our proposed technique.To successfully presume the posture of human head using the 3-D point cloud image of human face can help to make some improvements to the Safe Driving System and reduce the accidents.But ICP also has some problems. The main problem is the time of calculation is long. The main reason of it is the point number of the point cloud is very large. So the further work of us is to reduce the points of 3-D point cloud and in the same time successfully presume the posture of human head.