Full occlusion handling for pedestrian tracking via hybrid system

Abstract

Occlusion and lack of visibility even in sparse crowd scenes make it difficult to track individual pedestrians correctly and consistently, particularly in a single view. We present a novel pedestrian tracking approach that connects tracking with reidentification to locate and maintain the identity of certain people who may be occluded for a long time. First, two models are constructed. One model tracks the pedestrian and trains a classifier, while the other model reidentifies the pedestrian of interest from detection results with the trained classifier. Secondly, we design a set of transition rules for model switching. Finally, the two models work alternatively based on the principle of a hybrid system to track the pedestrian. Several typical sets of experiments show that the proposed approach outperforms the state-of-the-art approaches and achieves robust pedestrian tracking in the presence of full occlusion.