Real-time head pose estimation for mobile devices

Abstract

Mobile devices are an essential equipment in modern life. Nowadays, its presence is so widespread that almost everyone has a mobile phone, smartphone or tablet device. There are several different ways to interact with those equipments, such as the use of the keypad or the touchscreen. Here, we propose a real-time head pose estimation technique based on the secondary video camera as a means of interaction between the user and the device. The proposed technique is composed of several computer vision methods specially optimized to be able to operate in a restrict environment and a head pose estimation based on the calculations of the roll, yaw and pitch movements. Experiments were conducted based on 363 videos of 27 different people in a varied environment (illumination and background). © 2012 Springer-Verlag.