Emotion Evalution Of Moving Crowd Based On Fuzzy Reasoning

Crowd behavior analysis plays an important role not only in video surveillance,but also in public safety management.Generally,the crowd characteristics are extracted to analyze and judge the crowd status.This paper combines the knowledge of computer graphics,fuzzy reasoning,psychological emotion model,physics and other fields,and takes the crowd emotion evalution method based on fuzzy reasoning as the main research content.Firstly,the features related to crowd emotion are selected and extracted,and the Voronoi diagram is introduced to calculate the crowd density.Then,the fuzzy inference theory is introduced,the single feature input and double features input are used to construct the fuzzy inference system,respectively,to evaluate the crowd arousal and valence,and compare the outputs of each fuzzy system.The following is the innovation of this paper:1.Construction of emotion-related characteristics of the crowd.Among the numerous crowd characteristics,taking into account the relevant characteristics of crowd emotion and combining with the excitation-pleasure emotion model,the characteristics related to crowd motivation were found out: crowd enthalpy and variance of crowd motion amplitude,and the characteristics related to crowd pleasure were found out: crowd motion disorder index and crowd density.The four characteristics of the population as an important basis to obtain the mood of the crowd.2.In the field of video surveillance,a crowd density estimation method based on voronoi diagram is designed.Firstly,the LK optical flow tracking FAST feature points were used to calculate the motion feature points.Then,in order to reduce redundancy,only the most robust feature points are retained,which roughly represent the location of the crowd.Finally,the density of these feature points is obtained based on Voronoi diagram to represent the crowd density.Compared with other density estimation methods,this method focuses more on the consideration of local density and is more suitable for describing crowd emotions.3.Build a crowd emotion assessment system based on fuzzy reasoning.Firstly,the enthalpy and amplitude variance of the crowd are taken as the input of the fuzzy reasoning system of the degree of excitation,and the chaos index and density of the crowd are taken as the input of the fuzzy reasoning system of the degree of pleasure,and the degree of excitation and the degree of pleasure are fuzzy into the output.Then the fuzzy inference system based on single feature and double feature input is constructed respectively.In this paper,the single feature reasoning results and the dual characteristics of reasoning results comparing and artificial stated value,the results show that the selected crowd characteristics are correct,the method can also be roughly effective assessment group and pleasure degrees,at the same time,the characteristics of double input of fuzzy system is better than single character input of fuzzy system generalization performance.