岩溶石漠化是中国西南地区的一种严重的生态环境问题。近年来有进一步恶化的趋势。本文以贵州省普定县典型研究区IKONOS遥感数据为例，给出了一种利用多光谱遥感数据定量提取岩溶石漠化程度信息的算法。在多维光谱的向量空间分析的基础上，提出了“植被线”的概念，并定义“石漠化几何指数”（GRI）为多维向量空间中不同程度的石漠化像元到“植被线”的归一化距离。作者认为使用这种方法提取石漠化程度，可以避免人工解译的主观性强，工作量大的缺点。使提取到的石漠化程度结果客观稳定。

Rock-desertification is a unique environmental and ecological problem in Southwest China.As remote sensing is an important way of monitoring variation of rock-desertification,a method for measurement and information retrieval of rock-desertification is developed in this paper.MNF transform is applied to IKONOS data to reduce the number of spectral dimensions to 3.In this 3-demension,endmembers in the image are found out and analyzed.It is found that various vegetations group into a line called "vegetation line",in which "dark vegetations",such as coniferous forest and broadleaf forest,continuously change to "bright vegetations",such as grasses.It is presumed that is caused by deferent proportion of shadow mixed in leaves or branches in various types of vegetation.The normalized distance between points for rock-desertification and the vegetation line is defined as Geometrical Rock-desertification Index(GRI),which can be used to scale rock-desertification.This method was used in the IKONOS image of Puding,Guizhou province and showed many advantages.