A robust head pose estimation method based on POSIT algorithm

Estimating the head pose is an important ability of a computer when interacting with humans because the head pose usually indicates the focus of attention (Bennewit, Faber, Joho, Schreiber & Behnke 2005). In this paper, we present a novel method to estimate head pose over low-resolution web camera images. Our approach proceeds in three stages. First, a face detector localizes faces on the input image. Then, classifiers trained with AdaBoost using Haar-like features, detect distinctive facial features namely the mouth, the nose tip and the eyes. Based on the positions of these features, finally the POSIT algorithm estimates the three continuous rotation angles and the translation vector, what we use later fore head pose modeling. Since we have a compact representation in case of faces – using only few distinctive features –, so thus our approach is computationally highly efficient. As we show in experiments with standard databases as well as with real-time image data, our system locates the distinctive features with a high accuracy and provides robust estimating of head pose.