ISOMAP(Isometric Feature Mapping，等度规特征映射)算法在高维非线性数据处理中有较为理想的效果，建立在MDS (Multi-Dimensional Scaling，多维尺度变换)的基础上，其基本思想是当数据集的分布具有低维嵌入流形结构时，可以通过保距映射获得观测空间数据集在低维空间的表示。-ISOMAP (Isometric Feature Mapping, and other metric feature mapping) algorithm can get more satisfactory results in high-dimensional nonlinear data processing, based on the MDS (Multi-Dimensional Scaling, multidimensional scaling), based on the basic idea is that when the data set distribution of low dimensional embedding manifold structure, you can get insurance　the observation of spatial data sets mapped low dimensional space of representation .