Crowd panic state detection using entropy of the distribution of enthalpy

摘要

The motion state of a crowd can be described using the change of energy information of pedestrians' motion. Enthalpy can be used to describe the state of a system by consider the energy of the system. The distribution of enthalpy will change follow the change of a crowd state. Entropy is very suitable for measure the degree of disorder of a system Based on this idea, a crowd panic state detection method is proposed in this paper according to the entropy of the distribution of enthalpy. Firstly, the optical flow of two frames is calculated to get the motion information of a crowd. Secondly, based on the results of optical flow, the pedestrian moving region can be gained based on flow field visualization and texture segmentation method. Therefore the enthalpy in a tiny image region can be gained in the effective crowd movement region. The distribution of the enthalpy for the motion field with moving pedestrians can be gained. Based on the distribution of enthalpy, entropy of each frame can be calculated to describe the crowd state. Experimental results show the panic crowd motion state has higher entropy, and normal crowd state has lower entropy. (C) 2019 Elsevier B.V. All rights reserved.