A bus passenger flow estimation method based on feature point's trajectory clustering

Abstract

Based on the observation that motion of different pixels from the same target has very similar spatial-temporal properties in bus video surveillance images, a feature point's trajectory clustering method is proposed to estimate passenger flow in this paper. Firstly, the pyramid-based optical flow algorithm is utilized to tracking the feature point's movement in the images; then, their trajectories are pre-classified into passenger getting on, off the bus and others according their motion direction histogram; finally, the pre-classified trajectories are clustered by their spatial-temporal similarity and the cluster number is looked as the result of bus passenger flow estimation. Since it needn't to detect the head contour, face or other features of the passenger, our method is simple, fast and strong. The experiment results on multiple real bus surveillance videos show that it has high counting accuracy (>90%) in different illumination, background and even crowded conditions.