Coal fires detection study using adaptive-edge threshold algorithm

In order to detect positions of coal fires in Wuda coal field, Inner Mongolia, an adaptive-edge threshold algorithm (AETA) is proposed aiming at the thermal infrared band from domestic CBERS-04 remote sensing satellite.AETA first filters the noises of an image using Gaussian filter algorithm, generates high-temperature edge combined Sobel operator with extracted high-temperature regions, averages the pixel values located in high-temperature edge, and applies this mean of temperatures as temperature-segmented threshold to identify and extract coal fire areas.By comparing field-investigated fire areas with extracted fire areas, we find that the overlap ratio of both areas reaches 81.3%.Additionally, the two fire areas recognized by ASTER satellite and CBERS-04 satellite at night on the same day show the overlap ratio of both fire areas is 81.0%.These results demonstrate that AETA algorithm can provide the higher precision for coal fires detection.