Small fire smoke region location and recognition in satellite image

Abstract

Since ancient times, forest fires are an important threat to human beings, with the development of satellite technology, forest fires can be constantly monitored by satellite. But face to number of satellite pictures, its hard to detect fire location, especially the fire initial stage. An sub-region detection method has been proposed, this method supply fast partition domain detection method using information entropy, fast detect each sub-region whether there is smoke feature or not. Due to the satellite image has large region, our method first segment the image and then detect the smoke feature in each sub-region using information entropy. More than 50 pictures has been used to test this methods, shows this algorithm the effectiveness of small range fire detection.