Estimating Multi-Human Positions in Real Environment by Thermal Sensor

Abstract

This paper proposes a people locating method with room layout estimation by a thermal sensor. In the system, the sensor is attached to the ceiling and it acquires 16 × 16 elements spatial temperatures–thermal distribution. The distributions are analyzed to estimate people positions. Firstly, room temperature is removed from thermal distribution. Secondly, a distinction map including people positions is estimated with four fuzzy rules. In this procedure, an O-F (Object-Floor) map is calculated to verify people positions by temperature. The O-F map shows brief room layout and is employed to prevent miss detection of people positions. In the experiment, we measured a room to evaluate detection ability of our system. As the experimental result, the system successfully located people.