ETM+影像融合的评价分析

Evaluation and Analysis on Image Fusion of ETM+

针对Landsat-7 ETM+影像,对采用HIS变换、Brovey变换、相关系数加权法、SFIM变换、ISVR变换5种方法实现的ETM+全色波段与ETM+多光谱波段融合结果进行分析,并以均值、标准差、信息熵、平均梯度及相对偏差作为评价指标,通过试验对融合影像进行综合定量评价。结果表明,HIS变换光谱失真最大;Brovey变换在光谱保真和保持空间分辨率方面都不理想;相关系数加权法的融合结果严重地依赖权系数的取值;基于SFIM变换的融合方法,能最大限度地保留光谱特征,但空间分辨率提高不明显;而基于ISVR变换的融合方法,光谱失真小,空间分辨率高,其最大优势是考虑了多光谱影像与全色影像间的波段关系,可作为ETM+影像融合的最佳方法。

In this paper, five image fusion algorithms including HIS transformation,Brovey transformation, Weighting fusion, SFIM transformation and ISVR transformation are experimented between Landsat-7 ETM＋PAN band and multi-spectral bands. The result of different fusion methods is evaluated qualitatively and quantitatively by comparing the spectral fidelity and the spatial resolution. The study indicates that HIS transformation distorts the spectral quality most seriously. Brovey transformation is not very nice both in retaining the spectral quality and improving the spatial resolution. The result of Weighting fusion depends on the weighting seriously. Smoothing Filter-Based Intensity Modulation（SFIM） distorts the spectral quality greatly, however, it can not improve the spatial resolution. Improved Synthetic Variable Ratio（ISVR） is very fit for fusion with ETM＋ images on the whole. ISVR technique distorts the spectral quality slightly ,as well as improves the spatial resolution obviously. The advantage is that it considers the relation of panchromatic band and multispectral bands, which is ignored by other methods in this paper.