【摘要】 高山峡谷地区山高坡陡、地形破碎,大量的阴影及混合像元的存在,给遥感分类带来了困难。为了减少和消除这些因素的影响,采用分区分类方法,从NDVI中选择恰当的阈值,利用该值从原始图像上划分出植被区和非植被区并分别进行分类,试验结果表明,其分类精度(74.63%)较传统的直接分类精度(61.59%)有较大提高,是山区遥感图像分类行之有效的方法之一。更多还原

【Abstract】 The Dulong river is located in the mountain area with high elevation and steep slope,its topography is diverse.There are a lot of shadows and mixed pixels on the remote sensing image,which brings a lot of difficulty to the classification.In this paper a new method,sub-region classification,is proposed to deal with the classification of remote sensing image of mountain areas.In order to reduce and eliminate the affection of those factors,a sub-region method is proposed,i.e.selecting the right threshold of vegetation and nonvegetation regions from NDVI,then classifying vegetation and nonvegetation regions from remote sensing image using the threshold value.The result indicates that the classification accuracy(74.63%) is improved compared with the common method(61.59%).This method is one of the efficient methods in mountain area image classification.更多还原