高光谱遥感影像中最佳谱段的快速选择方法  

An Algorithm of Rapid Optimum Band Selection from Hyperspectral Remote Sensing Image

作　　者：尹继豪[1] 王义松[1] 

机构地区：[1]北京航空航天大学宇航学院,北京100191

出　　处：《遥感信息》2010年第3期3-6,12,共5页Remote Sensing Information

基　　金：航天科技创新基金资助项目(CASC200904);CAST创新基金资助项目(CAST200811)

摘　　要：针对高光谱遥感影像数据量大、信息冗余多的特点,在子空间划分理论基础上,结合最大熵和光谱角制图算法,提出一种快速的最佳谱段选择方法。充分利用相邻波段数据间的相关性分块特点,首先提取各子空间熵值最大的波段,然后依据地物光谱可分性选择最佳的波段组合。实验验证,最佳谱段选择速度快,并且所选谱段组合用于目标提取,效果显著。With the development of hyperspectral technology,the spectrum resolution is continuously improved.However it leads to a huge increase of image data and information redundancy.To solve the problem,the method of dimensionality reduction is discussed and some key algorithms are analyzed,such as entropy and joint entropy algorithm,optimum index factor algorithm and spectral angle mapper algorithm.By analyzing the effectiveness,limitation and computational complexity of various algorithms,a rapid optimum band selection method which is based on the theory of subspace partition is proposed against the shortage of existing algorithms.This method is an integration of maximum entropy and spectral angle mapper algorithm.It makes full use of the characteristics of sub-block between adjacent bands.We extract the band with the largest entropy of each sub-space firstly,and then select the best band combination based on the divisible features of spectral.By experiment,it confirms that the rapid optimum band selection method can extract targets effectively.

关 键 词：高光谱 子空间划分 熵 光谱角制图 谱段选择 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]