【摘要】 基于像素级的融合方法很多,本研究选择了同一地区的ETM+、SPOT5和QuickBird遥感数据,分别采用Pansharp、Gram-Schmidt、小波变换、主成分变换和比值变换五种常用的方法进行融合处理,并针对不同的融合影像,选择了信息熵、平均梯度、信噪比、偏差指数和均方根误差五个指标来对融合的结果进行客观评价,从而根据不同的数据源,选择最佳的融合方法。通过分析比较,得出不同的方法对空间分辨率相对较低的ETM+和SPOT5影响较小,而对高分辨率的QuickBird影响明显,并得出Pansharp的融合效果最理想。更多还原

【Abstract】 There are many pixel-level fusion methods for remote sensing images.ETM+,SPOT5 and QuickBird images of the same district were fused with different methods such as Pansharp,Gram-Schmidt,Wavelet,Principal and Brovey.According to the different fused images,5 indexes like entropy,average gradient,signal-to-noise ratio,deviation index and root mean square error were selected to evaluate the fused result,and to select the best fusion method from the different images.Analysis resulted that fusion methods had no great effect on lower resolution images such as the ETM+ and SPOT5 images,but significant influence on high resolution images like QuickBird.The result showed that Pansharp method had the best fused result.更多还原