大比例尺航片正射影像中的树冠提取  

Crown Extraction from Large-scale Aerial Orthophotos

作　　者：李超[1] 岳树峰[1] 王胜蕾[1] 王绪鹏[1] 刘银帮[1] 李凤日[1] 

机构地区：[1]东北林业大学,哈尔滨150040

出　　处：《东北林业大学学报》2010年第9期71-73,77,共4页Journal of Northeast Forestry University

基　　金：国家教育部本科生创新项目(091022503)

摘　　要：为使森林航片的分类精度达到单株林木提取水平,以凉水国家级自然保护区1:8000正射影像、DEM及红松(Pinus koraiensis)生境、立地条件特征为基础,对航片进行了传统的非监督分类、监督分类和专家分类,并对其分类机理进行了分析,提取了红松树冠光谱特征、纹理特征。在构建的红松树冠识别模型基础上,执行分类,并运用混淆矩阵分析法对分类结果进行了精度检验。结果表明:将纹理信息、DEM信息、红松生境及立地条件特征信息作为专家知识,构建红松树冠识别模型进行分类的方法,能有效地降低地物分类中"同物异谱"和"同谱异物"现象,使分类的总体精度达到96%,比监督分类提高了10%,从而使大比例尺森林航片分类精度达到了单株林木提取水平。In order to improve the classification accuracy of aerial photographs,spectral characteristics and texture features of Korean pine crown were extracted on the basis of 1：8000 orthophoto and digital elevation model （DEM） of Liangshui National Nature Reserve and the characteristics of habitat and site conditions of Korean pine.A set of crown image recognition models of Korean pine was constructed by analyzing the mechanism of the traditional non-supervised,supervised and expert classifications.After classification,the accuracy of pattern matching results was validated by the analytic method of confusion matrix.Results showed that the image recognition model of Korean pine could effectively reduce the phenomena of the same object with different spectrum and different objects with the same spectrum in classification.The total accuracy of this method was 96% and enhanced by 10%,compared with that of the supervised classification.The classification accuracy of large-scale forest aerial photographs reached the level of individual tree extraction.

关 键 词：正射影像 红松 监督分类 专家分类 树冠识别模型 

分 类 号：P237[天文地球—摄影测量与遥感] S791.247[天文地球—测绘科学与技术]