Spot5影像统计分析及最佳组合波段选择

遥感影像特征分析是影像融合和解译的基础,而对遥感数据源各个波段进行定量分析是图像融合的前提。本文以SPOT5为遥感数据信息源,在湖南省资兴市天鹅山林场进行了图像特性和波段组合的实验研究。研究主要采用了典型地物的光谱数据采集分析和遥感数据定量分析相结合的方法,计算出各波段之间的熵、相关系数和协方差。研究结果表明:SPOT5各波段的标准差大小顺序为:波段3>波段4>波段2>波段1,熵值的大小顺序则为:波段4>波段1>波段2>波段3;波段3与波段4的协方差最大,而且两波段又处在红外区域,说明红外波段之间独立性较强;从光谱辐射仪采集的数据来看,水体、草地、裸土地的光谱反射率就有很大的差异性。乔木树种的光谱反射率在可见光区域内非常接近,在红外———近红外区域内具有一定的差异,红外———近红外区域波谱对林业遥感研究具有极其重要的意义;利用联合熵和最佳指数方法确定了最佳波段组合为1(R)4(G)3(B)波段。

The Analysis of Statistical Characteristics of SPOT 5 Image and its Optimum Band Combination

Statistical characteristic analysis of remote sensing image is fundamental to image fusion and interpretation,and the quantitative analysis of each band is the premise of image fusion.In this paper,lots of experiments on band combination and statistical characteristic analysis have been studied with SPOT 5 in Tian'eshang forestry center,Zixing county,Hunan province.The method adopted was the combination of typical object spectrum and their image quantitative analysis,then entropy,correlation matrix,and covariance matrix between SPOT 5 bands were calculated.The results showed that: 1) band 3 has the biggest standard deviation,followed by band 4,2 and 1;2) band 4 has the biggest entropy,followed by band 1,2 and 3;3) band 2 and band 3 have the biggest covariance matrix,which means that the infrared bands are more independent.Meanwhile,the data collected from field spectrometry instrumentation showed that: 1) the spectrum reflectance is vary differently between water,sward and bare land;2) the spectrum reflectance of arbor is very adjacent in visible light,and has certain difference in the area from infrared to near infrared,which is very significant in the forestry remote sensing study.According to these results,the best band combination is band 1 for red,band 4 for green,and band 3 for blue,which is ascertained by the associated entropy and the optimum index factors.