Decomposition of pedestrian flow heatmap obtained with monitor-based tracking
Akinori Asahara · 2017
Indoor pedestrian tracks are now obtained with monitoring systems, such as video and LiDAR. Statistics using the track information, such as the number of pedestrians and average moving speeds, gives facility managers insights into how to improve the floor plan. However, the statistics sometimes convolute multiple modes of pedestrian movement trends such as “most pedestrians are going straight”. To address the problem, decomposition of the pedestrian-track statistics is proposed in this work. The proposed method is based on a mixed Markov-chain model for rough transition between areas. Decomposed statistics are calculated after classifying tracks with the model into several groups of pedestrians. Moreover, statistics decomposed with the proposed method and existing methods are experimentally demonstrated with actual pedestrian tracks in this paper.