基于光谱特征的城市人工地物分级分类方法研究

１９９７ 年６ 月利用ＰＨＩ推帚式成像光谱仪在北京市沙河镇进行了飞行实验，其主要目的是成像光谱技术应用于城市用地和建筑物分类中关键技术研究。对所获取的１５ 个波段的可见光－ 近红外数据进行了分析与处理，试图通过对地面覆盖物质的光谱响应特征的分析来进行城市人工目标的识别与分类。由于城市地物的光谱特征异常复杂，难于应用一般的模式分类算法分离出所有的类型，本文在对地面覆盖类型光谱响应分析的基础上，采用分层复合分类的方法，逐步将不同的地物类型分离开来，实验证明是非常有效的

Study on the Hierarchical Classification of Urban Manmade Objects Based on Spectral Features

An experiment fleight was made by pushbroom hyperspectral imager PHI in Shahe, Beijng during June, 1997. The main aim was to study the key technology of urban land use and construct classification by using hyperspectral technology. By analyzing the acquired 15 bands visible near infrared data, we try to classify the urban manmade objects through the analysis of the spectral signatures. Because of the complexity of the spectral features of urban manmade objects, it is difficult to recognize the diffrent classes by traditional classification algorithm. Based on the analysis of the spectral features, we proffer a strategy of classification that calssify the classes by composite algoritnm hierarchically. The result states itself that this method performs better than the familiar method.