Innovative geometric pose reconstruction for marker-based single camera tracking

Mobile augmented reality applications are in need of tracking systems which can be wearable and do not cause a high processing load, while still offering reasonable performance, robustness and accuracy. The motivation to develop yet another tracking algorithm is two-fold. Most of the existing approaches use classical optimization techniques such as the Gauss-Newton method. However, since those algorithms were developed to address general optimization problems, they do not fully exploit the structure of the pose estimation problem with its geometric constraint targets. Also, mixed reality applications demand that pose estimation be not only accurate but also robust and computationally efficient. Hence there is a need for algorithms that are as accurate as classical algorithms, yet are also globally convergent and fast enough for real-time applications. In this paper we introduce a new iterative geometric method for pose estimation from four co-planar points and we present the current status of PTrack, an infrared marker-based single camera tracking system benefiting from this approach. Our novel pose estimation algorithm identifies possible labels composed of retro-reflective markers in a 2D post-processing using a divide-and-conquer strategy to segment the camera's image space and attempts an iterative geometric 3D reconstruction of position and orientation in camera space. Tracking results are made available to applications through OpenTracker [OpenTracker 2006] framework. To analyse tracking accuracy and precision, we built a generic test-bed and compared PTrack to ARToolKit [Kato and Billinghurst 1999; Kato et al. 2000], one of the most wide-spread low-cost tracking solutions. Results show that our tracking system achieves competitive accuracy levels better than ARToolKit and close to commercial systems, while being highly stable and affordable.