A Network of Miniature Thermal Imaging Sensors for Object Detection and Tracking

This paper presents a concept and implementation of an infrared imaging sensor network for object localization and tracking. The sensor network uses multiple low-resolution (80× 80 pixels) microbolometric thermal cameras to detect, track and locate an object within the area of observation. The network uses information simultaneously acquired from multiple sensors to detect and extract additional information about object’s location. The use of thermal-imaging systems responsive to objects’ natural infrared radiation, makes the system resistant to external illumination and environmental conditions. At the same time, the use of infrared sensor requires application of specially designed, dedicated image processing techniques appropriate for this kind of sensor. The paper describes: image processing techniques, means of object localization, accuracy measurements, comparison to other known solutions and final conclusions.