While the Nintendo Wii and Microsoft Kinect are largely responsible for bringing gesture control into the lounge room, a new technology from OMRON promises to make gesture control mobile. Building on the Japanese company’s core “OKAO Vision” facial image sensing technology, OMRON has developed a new hand gesture recognition technology compatible not only with PCs, but also Android and iOS mobile operating systems.
Through a combination of extracting specific hand features from a large number of hand image samples and comparing the shape of a particular hand against a large amount of hand shape models, OMRON’s technology references a previously recorded camera image to simultaneously recognize the position, shape, and motion of a person’s hand. OMRON says this method enables speedy recognition of gestures while only using a small amount of memory, thus providing potential for the technology to be embedded in smartphones, tablets and other mobile devices.
The system can be triggered by a Kinect-like wave of the hand, or automatically by analyzing the relationship between the position and direction of the face and the shape or position of the hand. OMRON says this automatic triggering allows for more natural human-machine interaction.
With a captured image of a hand a minimum of 40 pixels high, the system can detect up, down, left and right motions, as well as various finger/hand shapes at distances ranging from around 10 cm (4 in) to several meters. Using a Snapdragon 1 GHz processor, the system can recognize gestures at up to 30 frames a second.
As well as the potential to bring gesture recognition capabilities to mobile devices, such as smartphones and tablets, OMRON says that because the technology works with webcams, devices with built-in cameras and dedicated digital cameras, it could also by used to control a PC, remotely trigger the shutter in a digital camera, or to control a TV.
While it’s unclear when we can expect to see applications or hardware built around the technology, aspects of it will be demonstrated at the Symposium on Sensing via Image Information (SSII 2012), being held in Japan from June 6 to 8.