Visual Receding Horizon Estimation for human presence detection

Damien Brulin (1) , Estelle Courtial (2) , Guillaume Allibert (1)

1 PRISME - Laboratoire pluridisciplinaire de recherche en ingénierie des systèmes, mécanique et énergétique
2 IRAUS - Département Images, Robotique, Automatique et Signal [Orléans]

This paper deals with human presence detection by using a receding horizon estimator based on computer vision results. The visual position estimation problem is formulated into a nonlinear constrained optimization problem in the image plane. A global model combining the behavior of the human motion and the camera model is used to estimate the evolution of the visual features on a past finite horizon. The main interest of this method is the capability to easily take into account constraints. Experimentations in two different configurations highlight the efficiency of the proposed approach, especially with an image occlusion treated as a visual constraint.

Mots clés: Receding Horizon Estimation; Human detection; Computer Vision

Domaines: Robotique [cs.RO]

Damien Brulin, Estelle Courtial, Guillaume Allibert. Visual Receding Horizon Estimation for human presence detection. IEEE International Conference on Robotics and Automation (ICRA 2009), IEEE, May 2009, Kobe, Japan. ⟨hal-01994114⟩