遥感影像融合方法的精度评价

Accuracy Evaluation of Remote Sensing Image Fusion Method

目前,高分辨率全色遥感影像和低空间分辨率的多光谱遥感影像融合是影像融合技术应用的主流。EN-VI 4.4遥感影像处理软件影像融合处理的工具——SPEAR提供了PCA变换、Gram-Schmid变换、Brovey变换和HSV变换4种专门用于全色与多光谱遥感影像融合的算法。以ETM+全色与多光谱遥感影像融合为例,选择熵、偏差和相关系数3个定量指标,对采用4种融合方法得到的影像进行评价。经综合评价和比较,实验的4种影像融合算法中,Gram-Schmidt变换效果最好。

ENVI（The Environment for Visualizing Images）is internationally well-known image processing system.ENVI version 4.4 affords a new image processing tool-SPEAR,which could wizard-style finish image fusion.The tool of SPEAR provide four fusion algorithms for the panchromatic and mufti-spectral images,that is,PCA（principal component analysis） transform,Gram-Schmid transformation,Brovey transformation and HSV transformation.In order to objectively and quantitatively evaluate precision of PCA,Gram-Schmid,Brovey and HSV transformation,four algorithm experiments were done in this paper based on ETM＋543band and panchromatic band.And three quantitative indicators were calculated and analyzed,that is,information entropy,correlation coefficient and deviation.According to comprehensive evaluation and comparison,the best effect is Gram-Schmid transformation,combined with fusion images through four transformation methods.