False alarm reduction in forest fires detection with low-cost commercial sensors

This paper presents an innovative system for false alarms reduction in automatic forest fires surveillance. The proposed algorithm is intended to operate on images sensed in Visible band. In particular, it reduce or eliminate false alarm come from Hot Spot Detection algorithm operating in the Thermal InfraRed (TIR) band. The false alarm reduction procedure, named Smoke Detection, is based on real-time integration in the model and image processing of images sensed in the Visible band. The algorithm is operating for test activities onboard the integrated monitoring system named SIRIO. Results of the validation of the system highlight how the algorithm improves the reliability of Hot Spot detection by eliminating a large number of false alarms.