论文部分内容阅读
引言划分区域边界(或分割)是遥感数字图象处理的基本问题之一。其方法大致有两类:一是直接寻求区域的边界;二是在区域分类的基础上实现分割。考虑到实际遥感图象在区域内部或边界上的复杂性,以及人们通常对于所研究的区域会有较多的先验知识,因此,我们着重研究建立在监督分类基础上的区域分割方法。在分类时原则上要充分利用遥感图象的多光谱信息以及实际类别在空间分布的结构特征,要求能够区分出多种类别。
Introduction Partitioning boundaries (or segmentation) is one of the basic problems of remote sensing digital image processing. There are basically two kinds of methods: one is to seek the boundary of the area directly; the other is to divide the area based on the classification. Considering the complexity of the actual remote sensing image in the region or the boundary, and the people usually have more prior knowledge of the region under study, we focus on the method of region segmentation based on the supervised classification. In principle, the multi-spectral information of remote sensing images and the structural features of the actual categories in the spatial distribution must be used in principle to require that multiple categories can be distinguished.