基于多光谱影像和专家决策法的作物分类研究.pdfVIP

, Journal ofAnhuiAgr.i Sci. 2011, 39( 25) : 15809- 15811 乔 李岩 1, 2 2 3 2 刘 磊 , 江 东 , 徐 敏 , 尹芳 ( 1. , 100081; 2. , 100101; 3. , 100029) [目的]探讨基于多光谱影像和专家决策法的作物分类, 验证 用单时相多光谱影像区分农作物的可行性[方法]以呼伦贝尔 地区典型农业种植区为研究区, 根据野外实测光谱数据, 寻找区分研究区主要作物大麦小麦油菜的最佳时间, 根据作物波谱特征, 采 用决策树方法, 结合光谱角度制图( SAM )等光谱匹配方法, 开展了作物分类研究[结果] 用 8月上旬获取的LandsatTM 影像, 在对影 像进行几何校正大气校正的基础上, 构建决策树, 成功提取了小麦大麦油菜种植草场的种植信息, 分类总体精度达到 86. 90% , K ap- pa系数达到 0. 831 1[结论]以典型时相的多光谱影像为数据源, 应用决策树方法提取作物类型信息, 具有较好的应用前景 遥感; 物候; 决策树; 作物类型 S 126 A 0517- 6611( 2011) 25- 15809- 03 Crops C lassification Based onM ulti-spectral Image andDecision TreeM ethod LIU Lei et al ( K e Laborator ofR esources Rem ote Sensing D igital Agr iculture, M inistr ofA griculture, Beijing 100081) Abstract [ Objective] The ami was to d iscuss the feasibilit of using single spectrum mi age to classif crops based on mult-i spectral mi age and scholarsdecision-making method. [M ethod] T aking the t pical agriculture plantation area in Hulunbeier area, according to field m eas- ured spectrum data, the optmi um tmi e ofm ain crops like barle , wheat and rapeseed were studied; based on crops spectrum characteristics, b d ing of decision-making treem ethod, and considering SAM spectralmatchingm ethod, classification of cropsw as studied. [ R esult] B dint of LandsatTM mi age gained in the first half ofAugust, based on geographic and atm osphericproof-reading, decision-m aking treew as construc- ted. P lantation information about wheat, barle , and rapeseed and plantation grassland w as ex tracted successfull . The general classification precision reached 86. 90% . Kappa coefficientw as 0. 831 1. [ Conclusion] T aking t pical spectrum mi age as data source, and appl ing dec-i sion-m akingm ethod to get crops t pes information had fine application future. Key words Remote sensing; Phenolog ; Decision tree