A System for Detecting Congestion using Optical Flow Approach

Abstract

Nowadays, millions of people travel internationally for mass gatherings that range from major sports events to fairs, festivals, concerts, railways, or political rallies. Such mass gatherings pose special risks for crowds, because large numbers of people in small areas can facilitate the spread of infectious diseases or increase the risk of injury which certainly leads to crowd stampedes. Crowd depends on several factors, such as age, mood, and consumption of drugs or drinks, will influence whether violence is harmful. Congested crowds are more likely to be violent. These issues should be addressed well for avoiding unusual situations in such places. Human movement activity is gaining increasing attention from current computer vision researchers. It is one of the chosen proactive research area for modern technical decades. The admiration is due to a large evolution of applications in surveillance, crowds and their dynamics. Because the process is of great scientific interest, it offers new computational challenges and because of a rapid increase in video surveillance technology deployed in public and private spaces. In this paper, we present a system for the detection and early warning of hazardous situations during huge gathering. It is based on optical flow computations and detects patterns of crowd movement that are characteristic for lethal congestions. For optical flow, Horn- Schunck’s method is embodied to compute the optical flow fields to the gathered video. Segmentation of video frames is done and optical flow is computed for individual segments. A threshold is set in such a way that the detection of congested region in video is identified easily through comparison with individual segments computed optical flow. Finally, we display the congested region for further preventive measures.