Thermal imaging and human detection

Abstract

With the rapid development, the current energy consumption of buildings is almost one-third of the country's total energy consumption. A smart energy management system is needed for the building to be intelligently controlled. The first aim for smart control is collecting data, which depends on the sensors. The sensors are chosen based on the parameters to be monitored. For example, smart control of Air Conditioner needs thermal sensors. In this project, we compared different sensors and the Grid-EYE thermopile sensor has been chosen. Then, calibration has been performed to the sensor to meet the requirements: human detection and moving direction judgement. The principle of human detection is pattern recognition. This article introduces the basic definition of pattern recognition, pattern recognition system and different methods of pattern recognition. After comparing different pattern recognition algorithms, the K-Nearest Neighbor algorithm has been chosen for the project. The experiment for human detection includes target judgement, people counting and direction determination. The sensor was placed at three different positions, and the results obtained were reasonably accurate.