Depth cameras aren't new technology, but shrinking them down to the size of a smartphone so that it can see in 3D the way a Kinect can remains a challenge. New work by researchers at Microsoft illustrates how any old smartphone could be used to sense 3D space.
Technology Review reports the new research which is being presented at SIGGRAPH today. The hope is to bring a simplified version of the 3D vision that's been tackled by Google's Project Tango on an ordinary phone or webcam. rather than using tons of application specific mobile devices with specialised chipsets and cameras, the research would enable 3D depth mapping with just a hardware tweak.
Of course, there's some assembly required. The hack calls for removing the infrared light filter that's used on most smartphones and webcams, and then adding a another filter, which blocks out everything but infrared light. Then you mount on a small ring attachment full of infrared LEDs, which act essentially like the infrared projector in a Kinect. That sounds complicated! According to the researchers, however, the hardware hack can be done in just 10 minutes. Since these are pros, you should expect it might take a little longer for you or me.
The research focused on using the infrared camera data to recognise hands and faces, sampling their movement at 220 frames per second. Software processing on the back-end matches a database of possible shapes to recognise the shapes in real-time. But with a larger database of shapes they could obviously do more.
The technology is obviously a little clunky for mainstream adoption right now, but with a little refinement, you could see how 3D depth mapping for some simpler gaming and scanning applications could totally come to a smartphone near you. Not that the tech behind Tango isn't impressive, but maybe people would be happier with slightly dumber tech they could strap to a phone of their choice. [Microsoft Research via Technology Review]