Direct 3D model-based tracking in omnidirectional images robust to large inter-frame motion

Seif-Eddine Guerbas (1) , Nathan Crombez (2) , Guillaume Caron (1, 3) , El Mustapha Mouaddib (1)

1 MIS - Modélisation, Information et Systèmes - UR UPJV 4290
2 CIAD - Connaissance et Intelligence Artificielle Distribuées [Dijon]
3 CNRS-AIST JRL - Joint Robotics Laboratory

This paper tackles direct 3D model-based pose tracking. It considers the Photometric Gaussian Mixtures (PGM) transform of omnidirectional images as direct features. The contributions include an adaptation of the pose optimization to omnidirectional cameras and a rethink of the initialization and optimization rules of the PGM extent. These enhancements produce a giant leap in the convergence domain width. Application to images acquired onboard a mobile robot within an urban environment described by a large 3D colored point cloud shows significant robustness to large inter-frame motion, compared to approaches that directly use pixel brightness as direct features.

Domaines: Automatique / Robotique; Vision par ordinateur et reconnaissance de formes [cs.CV]

Seif-Eddine Guerbas, Nathan Crombez, Guillaume Caron, El Mustapha Mouaddib. Direct 3D model-based tracking in omnidirectional images robust to large inter-frame motion. IEEE International Conference on Advanced Robotics, Dec 2021, Ljubjana, Slovenia. ⟨10.1109/ICAR53236.2021.9659324⟩. ⟨hal-03371352⟩