A new head pose estimating algorithm based on a novel feature space for driver assistant systems

Abstract

This paper introduces a Direct Accountability technique for real-time estimation of head's position and orientation. Unlike existing works which rely on feature extraction either in the image domain or in 3D space, our proposed approach based on expert classifiers ordering estimates the head pose for real-time applications, such as human-machine interaction (HMI) and video-based surveillance systems. In this paper a method for feature extraction is proposed which aims to keep only the angle related features that are independent of identity. Moreover a performance study is provided aiming to evaluate the accuracy of the proposed approach. Experimental results show that the proposed features could describe the pose of driver face image successfully, and have a good performance in real scenes specially for automobile safety and security and driver assistant devices.