Pedestrian Detection and Tracking Through Kalman Filtering

Abstract

There are a lot of challenges associated with - autonomous driving, and one such challenge is pedestrian detection and tracking, especially in this complex world, multiple people are involved in complicated and cluttered backgrounds. In this paper a novel method is proposed to detect, track and predict pedestrians based on Histograms of Oriented Gradients (HOG) algorithm and the Camshift algorithm respectively. These two algorithms run on top of the Kalman filtering framework. The Kalman filter is used as a tracker for precisely localizing and tracking the pedestrians. Then triangle similarity is used to calculate distance.