文献类型：期刊文献

中文题名：一种利用HJ-1B红外相机数据自动识别林火的方法

英文题名：An Automatic Forest Fires Identification Method Using HJ-1BIRS Data

作者：覃先林[1] 张子辉[2] 李增元[1]

第一作者：覃先林

机构：[1]中国林业科学研究院资源信息研究所,林业遥感与信息技术重点开放实验室,北京100091;[2]国家林业局森林防火预警监测信息中心,北京100714

年份：2010

期号：5

起止页码：700-706

中文期刊名：遥感技术与应用

外文期刊名：Remote Sensing Technology and Application

收录：CSTPCD;;北大核心:【北大核心2008】；CSCD:【CSCD_E2011_2012】；

基金：中国林业科学研究院科研基金项目(CAFYBB2007003);国家863计划项目;国家“十一五”科技计划课题(2006BAD23B04)共同资助

语种：中文

中文关键词：HJ-1B;红外多光谱成像仪;森林火灾;遥感

外文关键词：HJ-1B; IRS image; Forest fire; Remote sensing

分类号：TP79

摘要：森林火灾是一种世界性的重要自然灾害,它分布广、发生频度高,破坏森林资源,干扰人民正常生活秩序,造成全球性环境污染,越来越受到各国政府的重视。环境减灾1B卫星上的红外相机(简称HJ-1BIRS),其空间分辨率提高到了150m,高温饱和点达到了500K,是国内目前卫星上可探测地表温度最高的相机。在对HJ-1BIRS数据相关波段进行抽样统计分析基础上,针对HJ-1BIRS数据各波段特性,采用自适应的劈窗检测算法识别林火;在IDL语言环境下,实现了基于HJ-1BIRS和背景信息集成的林火自动识别算法程序。同时,通过近1a的试运行,并选取发生在东北林区和南方林区的森林火灾为验证案例,对算法及其监测精度进行验证。验证结果表明:该方法的判对率达到了90%以上,遗失率都低于10%,错判率为0,该方法基本能满足我国林火监测业务的精度要求。
Forest fire is a kind of worldwide natural calamity.It is extensively distributed with high occur-rence frequency and destroys forest resources thus disturbing normal living order of people and leading to environmental deterioration.The Infrared Sensor carried by HJ-1B（HJ-1BIRS）,its＇spatial resolution is 150mand saturation temperature is more than 500K.It＇s the only camera,which have been carried by Chi-nese satellite,to detect highest surface temperature in China at present.In this study,basing on analyzing the information of related bands of HJ-1BIRS by using sampling method,an adapted windows＇thread con-dition forest fire identification methodology has been developed according to the character of HJ-1BIRS.The automatic forest fire identification program has been developed by using HJ-1BIRS Image and integra-tion of background GIS data in IDL language.At the same time,by operating the program nearly one year,the fire identification method and identification results of precision have been validated by selecting the typ-ical forest fires which have been taken place in northeast or south forest region of China.The validation re-sults show that the truth precision is more than 90%,the missing identification precision low than 10%,but error is 0.So,the fire identification methodology can be satisfied the need for fire identification opera-tion in China.