Gestures are an important part of a range of mechanisms which we humans use to communicate with one another: they are a natural and instinctive mode of expression, which is in contrast to the way we currently communicate with technological devices, via keyboards, mice, dials and buttons. It is anticipated that gestures will, in the not too distant future, become an important mode of interaction with computers and many other diverse forms of technology.

For example, our hands are our natural pointing and signalling devices, yet, currently, we are limited to using keyboards, mice, touch-screens or, at best, cumbersome data-gloves for interaction with computers

Currently, we are working on methods which use Computer Vision techniques to build a reliable hand-gesture recognition system. We have build a simple prototype system which tracks a `region of interest' for the hand and then tries to classify the hand position into one of ten possible gestures.

The image below shows a sample screen-shot of the gesture-recogniser in action.

Main image Cyan-dashed-rectangle: bounding box of hand White-dashed-square: bounding square for hand; small-yellow-cross: centre of hand small-yellow-box: region for skin colour-sampling.

Inset images (top-to-bottom, left-to-right): matched example pose processed matched pose data data 'seen' as hand processed 'seen' image.