红外高光谱探测资料信息提取研究

项目名称： 红外高光谱探测资料信息提取研究

项目编号： No.41275029

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 天文学、地球科学

项目作者： 张水平

作者单位： 中国人民解放军理工大学

项目金额： 72万元

中文摘要： 本项目针对高光谱大气探测资料通道数量多、计算和存储量大、信息冗余影响反演或同化产品精度、在实际应用中需要进行信息提取的问题，从通道选择法和主分量特征提取法两种方法入手，进行信息提取研究。主要内容包括：1）对于通道选择法，探索以信息容量为指标，针对不同大气参数的反演或同化分别进行通道选择；2）对于主分量提取法，研究压缩后主分量所保留的信息容量从而给出确定主分量最优保留个数的方案，以及主分量压缩对观测噪音的降噪能力；3）针对不同大气参数的反演或同化，通过比较两种方法对某一参数的信息提取能力，给出分别适合于不同大气参数的优先信息提取方案。项目的完成有望为未来我国高光谱大气探测资料的有效应用提供技术支撑。

中文关键词： 红外高光谱探测资料；信息提取；主分量分析；通道选择；大气参数

英文摘要： Because of the higher spectral resolution and the greater number of channels, the infrared hyperspectral data has more information than the traditional infrared sounders, and the data processing method and atmospheric inversion method developed based on the traditional sounding data are not applicable to the infrared hyperspectral data. This paper gives a systematical research on the information distillation of hyperspectral data. The main contents of the project are as the following: 1) In channel selection, an approach of describing of the effective information of different atmospheric parameters contained in high spectral resolution atmospheric sounding data will be given in the project. And then a "Successive absorption" method for channel selection which using the information content as the optimal index will be designed. 2)The principal component analysis: with the information content as the standard of the compression, the distillation of hyperspectral data using the principal component analysis will be studied. And the noise reduction of hyperspectral data using the principal component analysis will carried out later. 3) For different atmospheric parameters, the information extraction ability of channel selection method and the principal component analysis method will be different. In the project, ba

英文关键词： infrared hyperspectral sounding data；information disillation；principal component analysis；channel selection；atmospheric parameter