Time-Of-Flight Technology Applied in Pedestrian Movement Detection

We propose a new method based on the Time-Of-Flight (TOF) measuring technology for the automatic pedestrian trajectory ex- traction. We start with a new data structure called TOF-tree for the segmentation of the frames in a sequence of recordings. The tree structure enables an efficient calculation of the segmented objects (TOF-objects). While a real object (pedestrian etc.) in the scene is always associated with a set of connecting TOF-objects, combined TOF-objects can be matched with each other from frames in the sequence to reconstruct the trajectories of the moving objects in the original scene. The trajectory information of the pedestrians can be adapted for industrial usage, for example, automatic passenger counting in public transportation.