3 of 4Image
The system works by using an off-the-shelf webcam to track the movements of the multicoloured glove and then uses software to map its movements onto a 3D computer-generated model, as seen here.
The glove is covered with 20 irregular-shaped patches, which allow the camera to recognise the front and back of the hand.
The patterns are painted in 10 different colours, composed of a palette chosen to maximise their contrast with the background and each other, and make them easier for the webcam to pick out.
The system, produced by students within the Computer Science and Artificial Intelligence Laboratory, is able to map even fine movements, such as pinching two fingers, on the model.
Photo credit: MIT
The system is powered by a software algorithm, developed at MIT, that is able to rapidly look up visual data in a database.
The software captures a picture of the hand, automatically crops out the background and then matches the picture against a database of myriad hand positions - all in a fraction of a second.
Aside from the obvious uses for gesture control to manipulate an operating system, researchers say the glove could be useful to engineers or designers looking for a more intuitive way to manipulate 3D computer-generated models of products or structures, as seen here.
Photo credit: MIT
Because the system relies on a database of relatively low-resolution 40 x 40px colour images it only takes up several hundred megabytes - small enough to be run on an average desktop computer, as seen here.
It only takes about three seconds for the user to calibrate the system, by pressing their hand on a piece of paper on a flat surface in front of the webcam.
Researchers within the computer lab are now investigating whether a pattern similar to the one used on the glove could be used on shirts to allow computers to capture information for whole body tracking.
You can see a video of the gesture recognition glove at work below.
Photo and video credit: MIT