Pedestrian tracking from a moving vehicle

Abstract

Intelligent vehicles and unattended driving systems of the future will need the ability to recognize rel- evant traffic participants (such as other vehicles, pedestrians, bicyclists, etc.) and detect dangerous situations ahead of time. An important component of these systems is one that is able to distinguish pedestrians and track their motion to make intel- ligent driving decisions. The associated computer vision problem that needs to be solved is detection and tracking of pedestrians from a moving cam- era, which is extremely challenging. Robust pedes- trian tracking performance can be achieved by tem- poral integration of the data in a probabilistic set- ting. We employ as hape model for pedestrians and an efficient variant of the Condensation tracker to achieve these objectives. The tracking is performed in the high-dimensional space of shape model pa- rameters which consists of Euclidean transforma- tion parameters and deformation parameters. Our Condensation tracker employs sampling on quasi- random points, improving its asymptotic complex- ity and robustness, and making it amenable to real- time implementation.