摘要
为了提高利用卫星图像编制土地覆盖分类图的精度，本研究提出将空间信息（结构特征量、分形特征量）叠加到光谱信息上（原图像），以提高分类精度的方法。利用ＳＰＯＴＨＲＶ图像对此法进行了检验，设立了９个探讨项目，对分类精度进行了比较。结果表明，用分形特征量进行分类时ＰＣＣ从８８．１％上升到了９０．２％。
The objective of this study is to evaluate the land cover classificationaccuracy for incorporating the effective spatial information such as texture and fra-ctal features into spectral information. It has been tested using SPOT HRV multi-spectral data.9 cases have been used to evaluate the approach and the classificationaccuracies have been compared.As a result,PCC(Probability of Correct Classifica-tion)is increased from 88.1% to 90.2% by incorporating fractal features.
出处
《环境遥感》
CSCD
1994年第2期150-160,T001,共12页
关键词
土地覆盖
分形特征量
分类图
精度
Texture feature,Fractal feature, Classification accuracy PCC
