摘要
多端元光谱混合分析是一种端元可变的线性光谱混合分析方法,通过由不同种类和数量的纯净像元(端元)构成的端元组合,对混合像元进行分解。针对每类地物,该方法可以采用多条同种端元光谱进行解混,在一定程度上克服了同种地物的光谱变异问题,能够提高解混的精度。本文对多端元光谱混合分析的具体方法进行综述。首先,基于对多端元解混研究现状的深入分析,归纳了多端元光谱混合分析的基本流程。其次,对多端元解混涉及的端元选取方法进行总结,分别概述了图像端元和参考端元扩充的策略及优化的指标;在此基础上,系统论述了端元光谱库构建主要途径及其优缺点,并指出针对特定的研究区域最佳端元模型确定的方法。最后,提出多端元光谱混合分析存在的问题并给出相应的解决方案。
Based on linear spectral mixture analysis,multiple endmember spectral mixture analysis is a method which considers the variability of endmember.Through a combination of spectral information and spatial information of pixels,this method can dynamically adjust the quantity and type of endmember pixel by pixel to complete the decomposition of every mixed pixel.It allows each pixel to be resolved with a variety of endmember models.When it comes to each type of ground object,sever spectra of each endmember can be used for spectral unmixing.To a certain degree,this method addresses the problem of spectral variation,which is caused by the reason that the same surface feature has different spectra.Therefore,this method represents a substantial improvement in accuracy over the traditional method which adopts a rigid set of endmembers to resolve mixed pixel.The multiple endmember spectral mixture analysis is reviewed in this paper.First of all,the necessity of analysing mixed pixel using numerous endmembers was discussed.Based on the full analysis of multiple endmember spectral mixture analysis,specific processes of this method were summarized.Then the image endmember extraction algorithms involved in the current method of multiple endmember spectral mixture analysis were also summarized.In addition,the approaches to expand and optimize the image endmembers and reference endmembers were concluded respectively in this paper.On this basis,the cardinal techniques of building endmember spectral library were studied systematically.Moreover,when it comes to specific research areas,the methods of selecting optimal endmember models were presented.Finally,the shortcomings that existed in the processes of multiple endmember spectral mixture analysis were summarized and the corresponding solutions to these problems were also offered.
出处
《遥感信息》
CSCD
北大核心
2016年第5期11-18,共8页
Remote Sensing Information
基金
国家自然科学基金(40971205
41671360)
关键词
多端元光谱混合分析
端元提取
端元扩充及优化
端元库构建
端元模型
multiple endmember spectral mixture analysis
endmember selection
endmember expansion and optimization
endmember library construction
endmember model
