Head pose recognition in advanced Driver Assistance System
Tzu-Chi Kao, Tsung-Ying Sun · 2017
Real-time head pose recognition is an important function of Advanced Driver Assistance System that need to consider the driver's intention. The majority of early head pose recognition techniques utilize multi-axis sensors or 2D camera images to estimate the head pose in 3D space. When running the head pose recognition systems, the driver has to carry a device with multi-axis sensors or the system needs preprocessing images which are taken from a 2D camera. In this paper, an algorithm is developed to utilize head pose Euler angles from the software development kit of the 3D camera (Intel RealSense SDK) to estimate driver's head pose and interested area without preprocessing images. Besides, the algorithm only uses a 3D camera (Intel RealSense F200) as the sensor without any wearable devices. The proposed system is evaluated by using the Intel Edison board to control different LEDs which presented which area driver is interested in. The experimental results showed that the system recognized the driver's interested area effectively.