Low power and economic consumption of energy are not normally used associated with image processing for gesture recognition. It requires not only low power hardware but also fast compact software. At the School of Electrical and Computer Engineering at Georgia Tech (Atlanta, USA) there is now a camera that conforms to these conditions: it combines new hardware and an efficient image processor, reducing its power consumption to the point where it can remain on, powered by a solar panel.

Other than a reduction in the number of frames per second which contributes to lower power use, the researchers, instead of comparing images pixel by pixel, improve the energy efficiency of the motion detection software thanks to a global detection, over the whole field of vision, of crucial details of the motion of the subject being followed. Instead of following changes on a large number of pixels, they reduce the frequency of measurements on the areas where they are analysing movements.

At first designed simply to reactivate peripherals from standby or sleep states, the process has proved capable of recognising specific gestures (a Z for example, drawn with an arm) and various patterns without a notable increase in power use. A camera with gesture recognition can now be powered with energy harvested on site, allowing it to be placed away from any power sources.

The first fields of application foreseen for this device are surveillance and robotics, but also in consumer electronics for specialised hands-free applications.