对Landsat TM/ETM+数据进行抽样统计分析基础上, 把Landsat TM/ETM+的红外、短波红外和近红外等波段数据相结合, 采用窗口动态阈值算法构建燃烧区识别模型；在此基础上, 为定量生成火线轮廓参数, 通过连通性判断、孔洞填充、小斑块去除和边缘细化等图像处理方法对识别的燃烧区进行处理；并在ENVI 4.8+IDL语言环境下, 实现了基于Landsat TM/ETM+数据自动生成火线轮廓参数算法处理过程的程序化.结果表明, 总体判对率为86.44%, 总体误判率为13.56%(其中漏判率为1.77%, 错判率为11.79%)；该方法可满足林火扑救中对火线轮廓参数定量宏观监测的应用需求.

To fulfil quantitative monitoring of fire line contour (location, length and area) for forest fire prevention and suppression, the algorithm for fire identification have been developed using dynamic thresholds of window combining the images obtained in thermal band, short wave infrared band and near infrared band of Landsat TM/ETM+. In order to get the quantitative result of fire line contour, the identified fire image has been processed by a series method such as the connectivity identification, hole filling, little speckle elimination and edge smooth. An automatic fire line contour identification program has also been developed using Landsat TM/ETM+images in ENVI 4.8+IDL condition. The method and program have been validated by selected typical forest fires which had taken place in northeast forest region of China and Амрская бласть of far east region of Russian. The validation results showed that the total accurate percent is 86.44%, the total false percent is 13.56% (the total omission is 1.77% and the commission is 11.79%). So, the methodology can be satisfied the need to knowing the fire line contour parameters for forest fire prevention and suppression operation.