迭代非监督分类快速提取林地信息的研究

Research on Iterative Unsupervised Classification of Rapid Extraction of Forest Information

迭代非监督分类是对传统非监督分类方法的改进,对传统非监督分类方法难以区分的部分进行反复提取和细化,最后将细化的分类部分插入到原非监督分类的图像中。在对大尺度区域的遥感图像植被分类工作中,迭代非监督分类方法是一种有效的方法。本文利用ENVI中的ISODSTA模块(迭代自组织数据分析算法),以哈密地区选取的实验区林地信息提取为应用实例,采用迭代非监督分类方法取得了令人满意的分类结果。

The traditional unsupervised classification method is improved by iterative unsupervised classification, which extracted and refined the distinguish part in traditional unsupervised classification method and insert the image classification refinement into the original unsupervised classification. The iterative unsupervised classification method is an effective method in remote sensing image classification of vegetation on large scale in the region. In this paper, forest information in Hami area from experimentation area is selected as an example; the ISODSTA module in ENVI （iterative self-organizing data analysis algorithm） and iterative unsupervised classification methods Were used, and achieved satisfactory classification results.