摘要
随着成像光谱仪的发展日益成熟,高光谱遥感图像的研究已进入到一个新的阶段--对获取的高光谱数据进行有效处理和利用的阶段。目前的处理方法主要集中在对高光谱图像的数值分析处理上,比如大气校正、降低数据维数、信息提取、分类与目标探测等方面。高光谱图像相邻的波段之间一般具有较大的相关性,并不是所有的波段对于后续处理都有着同等的重要性,通过选择最优波段而组成新的高光谱图像空间,在不损失重要信息的条件下可以代表其他波段的信息。因而需要在更合理的数学模型的指导下,按照一定的准则,来决定最佳波段的选择问题。对目前国内外高光谱图像的各种波段选择方法进行了综合归纳和分析,力图为高光谱图像处理寻找突破点,加强此领域的研究力度。
With the advanced development of spectrometer, research of hyperspectral remote sensing image has come to a new stage——processing and utilizing the acquired image efficiently. Now many processing methods are centered on arithmetic analysis processing of hyperspectral, such as atmosphere correction, dimensional reduction, information extraction, classification and target detection, etc. All bands don＇t have the same importance because there is great relation between adjacent bands, by selecting optimal bands, a new hyperspectral image space that can represent information of other bands can be assembled, under the condition that it will not lose important information. Therefore, under the guidance of more proper mathematical models, solving the problem of optimal band selection by observing some criterions. Various methods of band selection in domestic and foreign countries are generalized and analyzed, which tries to find breakthroughs and promote the research.
出处
《黑龙江大学自然科学学报》
CAS
北大核心
2007年第5期592-602,共11页
Journal of Natural Science of Heilongjiang University
基金
国家自然科学基金资助项目(60672034)
高等学校博士学科点基金项目(20060217021)
黑龙江省自然科学基金资助
关键词
高光谱遥感图像
降维
波段选择
hyperspectral remote sensing image
dimensional reduction
band selection
