以地物识别和分类为目标的高光谱数据挖掘

Hyperspectral Data Mining——Toward Target Recognition and Classification

高光谱信息挖掘技术是高光谱数据应用延拓与深入的重要环节，其核心在于光谱信息的挖掘。基于高光谱遥感信息的特点，探讨分析以地物识别与分类为目标的高光谱数据挖掘技术，包括基于模式识别的高光谱信息挖掘技术，基于光谱波形特征的挖掘技术，以及亚象元光谱信息挖掘。进一步发展有待于利用挖掘与知识发现的概念与模型，基于地物光谱特征及光谱数据库、知识库，在高光谱超维特征空间充分挖掘地物的光谱信息，以达到地物识别目的。

Hyperspectral Data Mining is a key technology to discover spectral information for application of hyperspectral remote sensing data. In this paper, focusing on target recognition and classification, we introduce some hyperspectral data mining algorithm, spectral feature\|based data mining algorithm, and sub\|pixel spectral information mining algorithm. In advance new algorithms should be developed to explore target spectral information in N \|dimensional feature space based on spectral knowledge and application modelling.