高分辨率影像香榧树分布信息提取  

Identification of distributional information Torreya Grandis Merrlllii using high resolution imagery

作　　者：韩凝[1] 张秀英[1] 王小明[1] 陈利苏[1] 王珂[1] 

机构地区：[1]浙江大学农业遥感与信息技术应用研究所,浙江杭州310029

出　　处：《浙江大学学报（工学版）》2010年第3期420-425,共6页Journal of Zhejiang University：Engineering Science

基　　金：国家自然科学基金资助项目(30671212);浙江省新苗人才计划资助项目(2007R40G2010015)

摘　　要：为了准确确定香榧树的空间分布、定量分析香榧树的适宜生长环境,基于IKONOS卫星影像,通过地统计半方差分析评价植被类型的可分性,并获取灰度共生矩阵纹理计算的最佳窗口;综合光谱信息、植被指数和纹理信息,应用C5.0决策树算法获取研究区地物分类的最优特征及规则,对香榧树的分布进行信息提取,其生产者精度为77.33%,用户精度为76.32%,该结果表明,基于决策树的香榧树分布遥感信息提取方法具有应用价值.In order to model the growth circumstance of Torreya Grandis Merrillii,obtaining its exact spatial distribution has significant importance. The present study identified the Torreya Grandis Merrillii distribution information from IKONOS imagery,combining the spectral and textural features. Semi-variograms were calculated to assess the separability of vegetation class and assess which spatial scales were most appropriate for calculation of grey-level co-occurrence texture measures to maximize structural class separation. Four spectral values,three vegetation indices and their grey-level co-occurrence texture measures were used in the decision tree model C5.0 to identify Torreya Grandis Merrillii. The producer＇s accuracy of identified Torreya Grandis Merrillii was 77.33%,and the user＇s accuracy was 76.32%,indicating that the proposed method is feasible to extract the distribution information of Torreya Grandis Merrillii.

关 键 词：香榧 植被指数 纹理 灰度共生矩阵 决策树 半方差分析 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]