CCTV analysis could predict and prevent disasters

September 13, 2007

The Swiss Federal Institute of Technology in Zurich has studied CCTV footage of disasters as it is believed that CCTV analysis could spot and warn of dangerous tensions to prevent them happening.

The team studied footage of the stampede on Saudi Arabia's Jamarat Bridge during the Hajj of 2006 and found that crowds can experience sudden changes like shock waves, turbulence, and even crowd quakes when built-up tension is suddenly released.

Dirk Helbing of the Swiss Federal Institute of Technology analysed the Jamarat Bridge footage along with Anders Johansson of Dresden University of Technology, Germany, and HE Habib Al-Abideen from Saudi Arabia's Central Directorate for Holy Areas Development.

The researchers used software to simplify the video and represent members of the crowd as moving patches of colour. They measured features such as the density, speed, and pressure of the crowd. The stampede occurred because too many people were funnelled into too small an area, but the team's analysis uncovered new features that might give forewarning of similar disasters.

Previous research suggested crowds move in smooth flows like a fluid, without sharp changes in direction. But, in this study, once the density of the crowd reached more than seven people per square metre this principle broke down.

Sharp compression waves moved through the crowd, shifting people back and forth, like grain or sand when driven through a funnel. During this period, which lasted 20 minutes, each person was alternately moving or stationary, and the waves of movement travelled through the crowd every 45 seconds.

This valuable work has indicated that in the future, organisers of events could use the software to analyse live CCTV footage and direct emergency services to areas where the tension is building up. Although preventing over-crowding in the first place is the best solution, the software could be applied to rock concerts, soccer games, or any big events.


Post a comment

Your email address will not be published. Required fields are marked *

Visited 533 times, 4 so far today