一种基于知识的金华地区土地遥感分层分类策略

提出了一种基于知识的分层次土地利用/土地覆盖分类策略,利用有限的基础资料,克服了图像时相不理想的问题,解决了解译难点,结合不同年度影像和相关背景资料,利用ErdasImagine的专家分类进行了遥感分类解译;第一层分类采用原有波段和仿植被指数,仿水体指数,仿城镇指数进行光谱分析并分类,第二层次将一次分类结果中混分的1992年水体城镇和2000年水体水田求交集获得两年度的水体分类结果,再由此求得1992年城镇和2000年水田分类结果;为基础资料有限条件下遥感解译提出了一个合理而有效的分类策略。

A knowledge-based hierachical classification for TM imagery of Jinhua

A knowledge-based hierachical classification strategy is proposed with limited and imperfect information through interpreter and classifying in Erdas Imagine.The classification of first level is completed by adopting original band and several index such as NDVI,WI,NDBI.The classification of second level gains the water area by computing the intersection of water &built-up area in 1992 and water &grain area in 2000 those can not classified well in the classification of first level.Then the built-up area in 1992,grain area in 2000 are computed by substration.This paper puts forward a reasonable and effective classification strategy of RS interpreter without enough information.