Tech-Note: Device-Free Interaction Spaces

Existing approaches to 3D input on wall-sized displays include tracking users with markers, using stereo- or depth-cameras or have users carry devices like the Nintendo Wiimote. Markers makes ad hoc usage difficult, and in public settings devices may easily get lost or stolen. Further, most camera-based approaches limit the area where users can interact.This paper presents Interaction Spaces - a distributed, optical sensor system for 3D input that lets users interact without needing markers or hand-held devices. An Interaction Space is created that covers the display wall. Inside it, objects like hands or fingers are tracked in 3D. This enables actions like moving or zooming a view on the wall. The added depth dimension allows images to be zoomed using a single hand instead of the two-hand "pinch" gesture used in other systems. The system's distributed aspect enables simple scaling to cover smaller or larger areas.The system is built using four computers and eight web cameras mounted along the floor. Each camera image is divided into vertical slices. Each slice is processed to detect ID object positions, before 2D positions are determined using triangulation. The 3D position of an object can be inferred from its corresponding 2D positions in each slice. The system is currently being used to control a microarray visualization on a 2x2 display wall. The system's accuracy has been evaluated, and is shown to be about I cm.