An Improved Forest Fire Monitoring Algorithm With Three-Dimensional Otsu

Forest fires can destroy millions of acres of land at shockingly fast speeds. The forest fire points identification algorithm is the most critical step in the forest fire monitoring process. Most traditional forest fire monitoring methods use fixed thresholds, ignoring background pixels, and have low recognition rates, which could lead to many problems, such as false reporting and low recognition rate. This paper proposes and tests an adaptive forest fire points identification algorithm using Himawari-8 data. By calculating the three-dimensional histogram of brightness temperature, an adaptive threshold that can automatically identify potential forest fire points is obtained. Based on this three-dimensional Otsu method, the contextual test algorithm has also been adopted to specify forest fire points. The experimental results show that the omission rate of the improved algorithm is about 10% lower than that of the previous algorithm in small-scale fire incidents. The improved algorithm can quickly and effectively extract fire point information, and it is also sensitive to small and low-temperature fires, which provides an efficient means for monitoring fire disasters.