NASA 1998 SBIR Phase I

PROPOSAL NUMBER: 98-1 25.03-6601

PROJECT TITLE: Using Natural Features to Track the Human Body in 3D


This is a proposal to develop a stereo vision based machine for full-body tracking of naturally clad humans in real-world environments. The proposed innovation will use naturally occurring visual texture (greyscale intensity variations) to simultaneously track all of the major segments of the human body in real-time. The foundation for this approach is the Proximity Space (PS) method, developed by Metrica researchers. The PS provides a means for efficiently packaging and controlling organized sets of correlation-based stereo and motion measurements in order to segment and track objects in 3d. For the purpose of tracking humans, the proposed system will use a kinematic model to constrain multiple PSs distributed among the primary segments of the body. Commercially available systems for tracking human pose are complex and expensive, requiring the user to encumber himself with numerous sensors, and/or sensing aids. Our approach avoids the constraints and expense imposed by artificial targets by capitalizing instead on the rich visual texture already available in the greyscale image of a human wearing ordinary street clothes. We expect that the robustness, ease of use, and low cost of this system will be met with much enthusiasm by developers and users of virtual reality and control environments.


The ultimate objective of this project is to produce a practical machine for accurate, real-time tracking of human pose (head, torso, and limbs). As envisioned, the proposed system would be portable and free-standing; a couple of binocular camera heads mounted on tripods wired to a Personal Computer. The market for human pose tracking systems, often referred to as "motion capture systems", is lucrative and rapidly expanding. Despite the current high costs of these systems, they are being used extensively by organizations involved in Virtual Reality, Tele-Operation, Biomechanics, Movie and Cartoon animation, and Robotics. Based on feedback from people who have used and tested more cumbersome and expensive technologies, we anticipate that this product will have a distinct competitive advantage.


Eric Huber
Metrica Inc.
1012 Hercules
Houston , TX 77058


Metrica Inc.
10010 San Pedro, Suite 400
San Antonio , TX 78216-3856