Mixed Scale Motion Recovery Using Guidable Cameras

 James Davis, Xing Chen

Abstract

Captured human motion is widely used by the medical, surveillance, and entertainment industries. In these fields, many interesting motions occur at mixed scales. Subtle localized motion often occurs together with long range movements. One of the chief challenges in recovering these motions is to obtain high resolution imagery suitable for resolving small details, while simultaneously increasing the working volume in which recovery is possible. This paper proposes an architecture for mixed scale motion recovery. Robust coverage of a large working volume is provided by a wide area tracking system. This system localizes interesting motions, and guides a separate foveated system of pan-tilt cameras to observe the detailed motion at high resolution. We demonstrate two applications, foveated structured light scanning and the capture of muscle deformation while walking. Both applications allow recovery of subtle detailed motion that would not be possible using existing single scale systems.

Selected Figures

High resolution imagers in the detailed tracking sub-system are guided to constantly observe the target as it moves around the entire working volume.

Top: Several frames from a captured video sequence of an object being moved in a volume of structured light. Bottom: The recovered depth map rendered from two viewpoints.

High resolution video of the subjects knee joint is captured from several views by our foveated motion capture system. The top two rows show synchronized video frames from two different viewpoints. The bottom row shows geometry recovered by triangulation. As a visual aid, the rendered geometry is shown superimposed on the video from camera 2.