Motion periodicity-based pedestrian detection and particle filter-based pedestrian tracking using stereo vision camera

Abstract

A methodology that detects harmonic motions of limbs and body during a typical human walk is presented. It temporally propagates the position, stride, direction and phase using a particle filter. This is based on a human limb-motion model, and is able to track the walking pedestrians in a heavily occluded environment. Potential 3D point clusters belonging to arms and feet are extracted employing an adapted version of RANSAC based surface detection algorithm. The periodicity feature is established via a Fourier-transform based periodogram that confirms the walk periodicity for each point-cluster representing limbs. RGB or intensity data from the stereo-vision input is completely ignored and the proposed method completely relies upon 3D data produced by the stereo-vision sensor. This independence from light-based information, produces reliable illumination invariant pedestrian detection and tracking results in outdoor environment using Daimler stereo pedestrian detection dataset.