TM遥感图像中居民点的自动提取

Extracting settlement on the TM image

介绍了一种TM遥感图像中居民点目标自动提取的方法.针对TM遥感图像的光谱特征,提出一种基于多波段信息的图像分割模型,提取出居民点、道路及河滩地等光谱相近的地物.然后依据居民点的形态特征分离道路,并利用空间关系知识分离出河滩地.试验结果表明该方法能快速准确地识别提取TM遥感图像中的大部分居民点.

In this paper, the method of extracting settlement from remote sensing image is discussed. Firstly, by analyzing the spectral features of settlement, it puts forward a simple settlement model based on multi-spectra to extract settlement, roads, bottomlands and other land-use types whose spectrum are close correlative. Secondly, it removes the roads by the shape features of settlement. Finally, the bottomlands are separated by spatial relation knowledge. The experimental results show that this method is effective in this field.