矿物高光谱解混进展研究综述  

Review of Research Progress on Mineral Hyperspectral Unmixing

作　　者：朱玲 李明 秦凯[1] 潘澄雨[1] ZHU Ling;LI Ming;QIN Kai;PAN Chengyu(National Key Laboratory of Remote Sensing Information and Imagery Analyzing Technology,Beijing Research Institute of Uranium Geology,Beijing 100029,China)

机构地区：[1]核工业北京地质研究院,遥感信息与图像分析技术国家级重点实验室,北京100029

出　　处：《遥感信息》2020年第3期15-23,共9页Remote Sensing Information

基　　金：国家自然科学基金青年基金项目(41602333)。

摘　　要：由于高光谱传感器低空间分辨率特征,岩石高光谱一般是矿物组分的综合反映。矿物高光谱解混对矿产勘查、矿物含量定量反演和野外地质填图等提供了可行的鉴定方法。首先介绍了2种主要的光谱混合模型;其次基于矿物混合机理特征,从模型驱动法和数据驱动法2个方面,对近年高光谱数据的端元提取和丰度求解算法进行归纳,分析各解混算法的原理和优缺点;然后从实验室实测数据、模拟数据和高光谱影像数据3个方面,对目前已开展的混合矿物高光谱解混实验进行概括,总结各算法的解混效果和适用性;最后针对各解混算法的特点和研究现状指出未来矿物高光谱解混的研究方向。Due to the low spatial resolution of hyperspectral sensors,the rock spectrum acquired by hyperspectral sensors is generally the mixing spectrum of mineral components.Mineral hyperspectral unmixing is of great significance on mineral exploration,quantitative inversion of mineral components and geological mapping.In this research,firstly,two main mixing models are systematically introduced.Secondly,based on the characteristics of the mineral mixing mechanism,the endmember extraction and abundance fraction estimation algorithms of hyperspectral data in recent years are summarized from two aspects of model-driven method and data-driven method.The basic theory is summarized and the advantages and disadvantages of each unmixing algorithm are also analyzed.Thirdly,the experiments of mineral spectral unmixing are summarized from three aspects of laboratory measured data,simulated data and real hyperspectral data.The unmixing accuracy and applicability of each algorithm are summarized.Finally,the future research direction of mineral hyperspectral unmixing is pointed out.

关 键 词：矿物高光谱 混合模型 解混算法 模型驱动 数据驱动 端元提取 丰度求解 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]