Research On Forest Fire Detection Technology Based On Video Image

Forest fire, as a natural and social phenomenon, brings a great threat to mankind. It's very important to detect forest fires early and accurately, so as to wipe them up when they are small and early, avoide the losses caused by the fires.The article proceeds from the characteristics of forest fire and the actual situation of video image forest fire detection, describes the detection technology of video image, and builds the analysis platform using VS2008 and OpenCV. Selects and products representative sample dates from a large number of collected forest fires, and uses the platform to verificate the four forest fires detection methods, that's background subtraction method, knowledge of forest fires method, frame subtraction method and synthesis method. Studies the threshold of background subtraction method and frame subtraction method; the criterion of knowledge of forest fires method comes after the fixed color space(HSI color space) and is derived from the statistical data, including day forest fire, night forest fire and day forest smoke. In addition, the article removes the moon lights, residents'lights and other interference by deducting some geographical position and spatiallocation, so as to decrease the false rate of arithmetics. The experiments show that these forest fire detection methods can effectively detect the experimental forest fires video, providing a high accuracy rate. Finally, the article prospects some future work to be done in future, providing a research direction for forest fire detection.