应用光谱混合分析法从SPOT影像提取槟榔树专题信息

台湾地区槟榔树分布面积较大,但是种植密度不高,用通常的遥感影像分类和专题信息提取方法不能有效地提取槟榔树的分布范围。光谱混合分析是将一个像元看成多种地物的组合,分别求出各种地物所占的比例。运用线性光谱混合分析的方法,对台湾的SPOT卫星影像进行像元分解,提出了如何利用低光谱分辨率影像分解像元的方法。通过与最大似然分类和阈值法提取结果的比较,表明光谱混合分析法不仅有效地提取了槟榔树分布的范围,还计算出槟榔树分布的密度,是一种较好的方法。

Application of Linear Spectral Mixture Analysis in Extracting Areca Thematic Information from SPOT Image

There are many areca trees planted in Taiwan for economic reason. But the areca will do great harm to environment and people's health, and it will lead to slope erosion for its shallow root, so the planting of areca must be controlled. Remote sensing provides a useful means to monitoring the planted area of areca. But the planted densities of areca are usually low, it is difficult to identify the arecas by ordinary methods even in high ground resolution remote sensing images such as SPOT. Linear spectral mixture analysis is a method to look a pixel as composed of many different kinds of endnumbers, and it can calculate the proportion of each kind of endnumber in each pixel, so it is useful to identify the small objects in the images. In this paper, linear spectral mixture analysis is used to calculate the density of areca. Some countermeasures were put forward to make up the shortcoming of less spectral resolution of SPOT images. Then the result of linear spectral mixture analysis was compared with those of maximum likeness classification and extraction by threshold. It can be found that the result from linear spectral mixture analysis can not only extract the correct area of areca, but also calculate its density, and it is much better than those of other two methods.