Real-time 3D pose estimation from monocular image sequencesis a challenging research topic. Although current methods areable to recover 3D pose, they are severely challenged by the computationalcost. To address this problem, we propose a tracking and 3D poseestimation method supported by three main pillars: a pyramidal structure,an aspect graph and the checkpoints. Once initialized the systemsperforms a top-down tracking. At a high level it detects the position ofthe object and segments its time-space trajectory. This stage increasesthe stability and the robustness for the tracking process. Our main objectiveis the 3D pose estimation, the pose is estimated only in relevantevents of the segmented trajectory, which reduces the computational effortrequired. In order to obtain the 3D pose estimation in the completetrajectory, an interpolation method, based on the aspect graph describingthe structure of the object´s surface, can be used to roughly estimate theposes between two relevant events. This early version of the method hasbeen developed to work with a specific type of polyhedron with strong edges, texture and differentiated faces, a die.


Torres, F., & Kropatsch, W. (2012). Top-Down Tracking and Estimating 3D Pose of a Die. In G. Gimel’farb, E. Hancock, A. Imiya, A. Kuijper, M. Kudo, S. Omachi, T. Windeatt, & K. Yamada (Eds.), Joint IAPR International Workshops on Statistical Techniques in Pattern Recognition (SPR) and Structural and Syntactic Pattern Recognition (SSPR) (pp. 492–500). Springer 2012. https://doi.org/10.1007/978-3-642-34166-3_54