Near-infrared composite pattern projection for continuous motion hand-computer interaction

Traditional structured light techniques require multiple patterns to be projected onto a scene and captured for three-dimensional (3D) reconstruction without ambiguity and resistance to variation of albedo. However, multi-pattern techniques can be corrupted by the object movement during the sequential projection. We have introduced a single projected pattern which efficiently combines multiple patterns into a single composite pattern projection-allowing for video rate three-dimensional data acquisition. Attaining low cost 3D video acquisition would have a profound impact on applications such as human-computer interaction and assistive technology. So far the composite pattern technique has only been performed with visible light which would cause user annoyances for these applications. To solve the problem, we use near-infrared to project the pattern. But this raises another more general issue of spatial skin response. In this study, near-infrared illumination is used for imperceptible measurement of hand pose. In particular, we are studying continuous motion depth acquisition for tracking hand motion and rotation as an interface to a virtual reality. Applications include advanced cockpit controls and computer interfacing for the disabled.