Using a depth camera as a touch sensor

We explore the application of depth-sensing cameras to detect touch on a tabletop. Limits of depth estimate resolution and line of sight requirements dictate that the determination of the moment of touch will not be as precise as that of more direct sensing techniques such as capacitive touch screens. However, using a depth-sensing camera to detect touch has significant advantages: first, the interactive surface need not be instrumented. Secondly, this approach allows touch sensing on non-flat surfaces. Finally, information about the shape of the users and their arms and hands above the surface may be exploited in useful ways, such as determining hover state, or that multiple touches are from same hand or from the same user. We present techniques and findings using Microsoft Kinect.