Thermal imaging tracks head count

Personal safety in football grounds and similar places could be improved by a new generation of people-counting thermal imagers described at the Institute of Physics conference on Sensors and their Applications.

Personal safety in nightclubs, football grounds and similar places could be improved by a new generation of people-counting thermal imagers described by Dr. Steve Hollock at the Institute of Physics conference in London on Sensors and their Applications.

The sensor that has been developed by IRISYS (InfraRed Integrated Systems Ltd) will bridge the gap between accurate but expensive thermal imagers with 50,000 pixels and the simple, cheap infrared-beam sensors that are too primitive for many desired applications.

Closed circuit television is often used to monitor people going into and out of public places, but there are many problems in using this kind of system if an accurate count of people is needed.

It is a very time-consuming process if an individual has to do the count, and automatic systems are often confused when the light isn’t good, or the same people wander backward and forward through the field of view. Another kind of sensor – an active infrared beam – is also problematic, as two people cutting the beam at the same time will only register as one.

The detector developed by Dr Hollock and colleagues is said to count the number of people in the field of view irrespective of the lighting conditions. It can even find out the speed and direction that someone is moving, their size and the length of time they have been in the field of view.

By linking up a number of such devices, it is also possible to get an overall picture of the motion of people, eliminating counting the same individual twice, or even tracking someone as they go about their business.

This new sensor has a detector array made up of 256 (16 by 16) elements formed by a pattern of electrodes on the surface of a ceramic chip. It is connected to a silicon chip that controls the output of the detector elements and allows the data to be processed.

A viewing window is attached to the detector and has a filter to select the wavelength of radiation that falls on the detector elements.

Human bodies radiate most intensely around 10 microns, so to detect people, a filter is chosen to let this radiation through. Also, the key to making this detector simple is that the detector only responds to changes in temperature, so it does not ‘see’ individual static objects in the scene, which is known as scene clutter.

By only giving a signal when an object moves within the scene, this detector reportedly saves a lot of computing power compared with video based systems. The accuracy of the detector is extremely accurate, as it can resolve up to a 0.5-degree centigrade change in temperature.