Mining Multi Human Locations Using Thermopile Array Sensors

This paper proposes multi human location estimation system using four thermopile array sensors. These sensors are attached to the ceiling and acquire 8 × 8 place-dependent thermal distributions. Firstly, human area is detected by background removal with fuzzy inference based on human characteristics. Secondly, human areas are identified by Connected Component Labeling. Thirdly, label is checked to detect adjoining people. Finally, the number of people and their locations are estimated as the labels. In our experiment, we employed four adult persons and they performed four motion patterns. As the result, the system estimated the peoples successfully.