Vision-based forest fire detection in aerial images for firefighting using UAVs

Abstract

Due to their rapid maneuverability and improved personnel safety, unmanned aerial vehicles (UAVs) with vision-based systems have great potential for monitoring and detecting forest fires. In this paper, a novel forest fire detection method utilizing both color and motion features is described for UAV-based forest firefighting applications. First, a color decision rule is designed to extract fire-colored pixels as fire candidate regions by making use of chromatic feature of fire. Then, the Horn and Schunck optical flow algorithm is employed to compute motion vectors of the candidate regions. The motion feature is also estimated from the optical flow results to distinguish fire from other fire analogues. Through thresholding and performing morphological operations on the motion vectors, binary images are then obtained. Finally, fires are located in each binary image using the blob counter method. Experiments are conducted, and the experimental results validate that the proposed method can effectively extract and track fire pixels in aerial video sequences. Good performance is expected to significantly improve the accuracy of fire detection and reduce false alarm rates.