抄録
Posture estimation of the human head is very important in various fields, such as safe driving. In the past, the facial features of 2-D color image are often used. However, the presumed accuracy is influenced by environment light. In our study, we use a device called Kinect to get the 3-D data of human head. Then we compare the head’s data with the one looking to the front to calculate the value of rotation. In particular, we rotate and translate the head many times until they are in the same direction. Finally, we calculate the value of rotation and translation, and presume the posture.
