Forest Fire Detection with Color Features and Wavelet Analysis Based on Aerial Imagery

Abstract

Unmanned aerial vehicles (UAVs), equipped with vision-based systems, can be used for forest fire monitoring and detection due to their low cost, fast response capability, and high safety. This paper proposes a novel approach to forest fire detection, which uses the color characteristics of the images taken by the UAVs and uses wavelet analysis to further process. Firstly, according to the color characteristics of forest flame and smoke, a low computational cost algorithm is adopted to extract pixels from its related regions. In order to correct the inaccuracy of color feature extraction, a two-dimensional discrete wavelet transform (nWT) is implemented to distinguish flame and the smoke area from other high-frequency noise signals. Multiple sets of experiments have proved that the algorithm proposed can effectively detect the forest flame and smoke part of the image. The good performance is anticipated to significantly improve the accuracy of forest fire detection on the basis of less computational cost and can perform real-time detection on the UAVs platform.