Monocular Head Tracking for Desktop Virtual Environments

Abstract

We present an approach to head tracking to be applied in desktop virtual environments. The system is able to estimate both the position and orientation of the user s head using only monocular view. In our approach, the edge of the front face of shutter or polarization glasses is detected, and the pose is estimated using algorithm based on planar homographies. The solution is based on marker configuration made up from lines, whose intersection points are used to estimate the pose. Instead of using planar square markers common in augmented reality applications, we take the advantage of the natural shape of the virtual reality glasses. In this paper, we describe our system set-up and detail the steps to implement the algorithm. In addition, we compare the proposed approach to well-known solution. The system performs real-time in a standard laptop computer.