High resolution, low cost, privacy preserving human motion tracking system via passive thermal sensing

Abstract

Thermal imaging is powerful but expensive. This thesis presents an alternative thermal sensing system capable of tracking human motion by using a novel projection mechanism from an array of inexpensive single-bit thermal sensors. Since thermal sensor performance is currently limited by lengthy refractory periods, the motion tracking system is further developed in the visible spectrum to demonstrate a proof of concept. A single array of sensors is coupled with Gray-coded binary masks to spatially divide a room, allowing for the tracking of a moving persons location. By analyzing the output over time, a motion flow map is rendered and displayed in a software graphical user interface. The software is also capable of preliminary two-person motion tracking schemes. This overall system provides a low-cost alternative for privacy preserving activity monitoring, and generating alerts based on data anomalies, such as a break in motion pattern or an unusually low level of overall activity. Future hardware and software advancements will allow a full-scale motion tracking system to be developed.