摘要
WorldView-2能够提供1个0.5 m全色波段和8个1.8 m分辨率的多光谱波段,为用户提供进行精确变化检测和制图的能力。本研究以深圳市植被为例,采用WorldView-2高分辨率遥感影像为数据源,进行缨帽变换及主成分分析处理,利用决策树分类模型进行提取。结果表明:WorldView-2影像经过缨帽变换及主成分分析处理后,能够明显增强影像的纹理信息,突出地物特征,并以各地物在经过缨帽变换及主成分分析处理之后的灰度值作为决策树分类的阈值,分类的总体精度、Kappa系数分别为89.26%、0.87,与以往的只利用波段的灰度值及植被指数等作为阈值相比,精度明显提高,方法也得到改善,得到了比较好的分类结果。
WorldView-2 can provide one band that the panchromatic band is 0.5 meters,and eight resolution multispectral bands that the panchromatic bands all are 1.8 meters,and have the ability for users to provide accurate change detection and mapping. With vegetation in Shenzhen city as an example,by adopting the Worldview-2 high resolution remote sensing images as data sources,the Tasseled Cap Transformation and Principal Components Analysis treatment for the vegetation information of Shenzhen city were conducted,further the data extraction was carried out by using decision tree classifi cation model. The results show that having been deal with Tasseled Cap Transformation and Principal Component Analysis,the image texture information was signifi cantly enhanced,the characteristics of ground objects were highlighted; by taking the gray values of all ground objects that were treated with Tasseled Cap Transformation and Principal Component Analysis as the decision tree classifi cation thresholds,the overall precision of plant classifi cation and the Kappa coeffi cient were calculated,being respectively 89.26% and 0.87,compared with the threshold only with previous bands of gray values and vegetation index,the new methods obviously improve the precision,the method is improved,better classifi cation results were obtained.
出处
《中南林业科技大学学报》
CAS
CSCD
北大核心
2014年第6期81-84,共4页
Journal of Central South University of Forestry & Technology
基金
"十二五"国家高技术研究发展计划(863计划)课题:"数字化森林资源监测关键技术研究"(2012AA102001)
国家自然科学基金项目"森林资源遥感监测波段窗口研究"(31370639)资助
