Forest Fire Detection and Localization Using Thermal and Visual Cameras

摘要

In this paper, a system targeting for using unmanned aerial vehicle (UAV) to application of detecting and locating forest fires is developed. It utilizes the information of two cameras: one charged coupled device (CCD) camera and one thermal camera. The information of both cameras are aligned and then fused to verify if a fire has occurred. In the alignment process, a homography matrix is calculated to assure that the corresponding pixels in both images point to the same area. An image-based thermo-visual servoing (IBTVS) system is implemented for the fire tracking process. A two-degree-of-freedom (2DOF) frame is fabricated based on IBTVS to test the tracking and locating capabilities of the system. Experimental tests show that the designed system can extract fire features from thermal and visual cameras, calculate the position of fire, and command the frame for pointing to the fire.