Pedestrian detection using sparse Gabor filter and support vector machine

Abstract

Vehicle warning and control systems are the key component of ITS. Pedestrian detection is an important research content of vehicle active safety. The central idea behind such pedestrian safety systems is to protect the pedestrian from injuries. In this paper, we address the problem of pedestrian represent and detection where the motion cue is not used. Inspired by the work proposed by Zehang Sun [2004], we proposed a pedestrian feature representation approach based on sparse Gabor filters (SGF) learning from examples. In the phase of pedestrian detection, we used support vector machine to detect the pedestrian. Promising results demonstrate the potential of the proposed framework.