不同组分比例的高光谱混合象元反射光谱形态特征研究  

Study on Reflectance Spectra Morphological Character of the Hyperspectral Mixed Pixels at Different Component Proportion

作　　者：朱锋[1] 宫辉力[1] 孙天琳[2] 侯婕[1] 郭小萌[1] 郭琳[1] 

机构地区：[1]首都师范大学资源环境与旅游学院,北京100048 [2]东北师范大学城市与环境科学学院,吉林长春130024

出　　处：《光谱学与光谱分析》2013年第7期1897-1902,共6页Spectroscopy and Spectral Analysis

基　　金：国家自然科学基金项目(41171335/D010702);国家自然科学基金重点项目(41130744/D0107);国家(973计划)前期研究专项(2012CB723403);北京市自然科学基金重点项目(8101002)资助

摘　　要：实验条件下,利用地物多角度二向反射平台及ASD地物波谱仪获取了由清水与荷叶组成的高光谱混合象元反射光谱,通过选择反射光谱形态指标,分析了24组不同组分比例下高光谱混合象元反射光谱形态指标变化规律,并得到以下结论:选择的光谱形态指标之间存在着较大的相关性,按照0.05及以上置信水平和可决系数R2>0.5的条件对光谱形态指标进行剔除,确定了最能体现反射光谱特征的形态指标;不同波段范围内,各光谱形态指标随混合象元成分比例的变化规律有所差异;聚类分析结果表明,纯象元(清水)过渡到混合象元存在临界值,当混合象元组分比例小于此临界值时,主要体现纯象元反射光谱特征,大于临界值时才表现混合象元的反射光谱特征。Under the experimental condition, the hyperspectral reflectance spectra of the mixed pixel of lotus and water body were acquired using the multi-angles dichotie reflex platform and ASD spectrum instrument. The reflectance spectral morpholog- ical character index variation of the hyperspectral mixed pixel at different component proportion was analyzed. The conclusions are as follows： the correlation between these morphological indexes was obvious, and the best morphological index was chosen by removing the index whose Rz was greater than 0. 5 （R is correlation coefficient） ; the variation of different morphological index was obvious; the cluster analysis shows that there was a critical value in the progress of the transformation from the pure pixel （water） to mixed pixel. If the proportion was less than the critical value, the reflectance spectra reflected the reflection charac- teristics of pure pixel; otherwise, it was the reflection of characteristics of mixed pixel.

关 键 词：混合象元 反射光谱 形态指标 聚类 

分 类 号：TP7[自动化与计算机技术—检测技术与自动化装置]