Rapid Hand Shape Reconstruction with Chebyshev Phase Shifting

Abstract

Human hand motion and shape sensing is an area of high interest in medical communities and for human interaction researchers. Measurement of small hand movements could help professionals to quantize the stage of conditions like Parkinson's Disease (PD) and Essential Tremor (ET). Similar data is also useful for designers of human interaction algorithms to infer information about hand pose and gesture recognition. In this paper we present a structured light sensor capable of measuring hand shape and color at 121 FPS. Our algorithm uses a novel structured light method developed by us, called Chebyshev Phase Shifting (CPS). This method uses a digital projector and a camera to create high-resolution color 3D models from sequences of color images. We show how to encode CPS patterns in three RGB images for a reduced acquisition time, enabling high speed capture. We have built a prototype to measure rapid trembling hands. Our results show our prototype accurately captures fast tremors similar to those of PD patients. Color 3D model sequences recorded at high speed with our sensor will be used to study hand kinematic properties in a future.