国产高分辨率遥感影像融合方法比较与分析

Research on fusion method comparison and analysis for domestic high resolution satellite images

GF-1,ZY-3和ZY1 02C三种国产卫星数据为研究对象,在分析单波段信息熵的基础上,运用联合熵,最佳指数和相关系数矩阵确定3种影像数据的最佳波段组合,选择具代表性的PC、Brovey、GS、IHS和MT变换融合等五种基于像素的融合方法对影像进行融合,分析融合前后影像联合熵以及融合后各影像波段的梯度变化。研究结果表明:(1)GF-1的最佳组合为Band1、Band3及Band4,ZY-3的最佳组合为Band2、Band3及Band4,ZY-1 02C数据的最佳波段组合为Band1、Band2及Band3;(2)融合后影像中Brovey变换融合的联合熵小于融合前,其余4种融合方法联合熵均大于融合前联合熵值;(3)GF-1、ZY-3和ZY 1 02C影像数据最佳融合方法分别为GS融合、PC融合以及IHS变换融合。

Three kinds of domestic high resolution satellite images such as GF-1, ZY-3 and ZY1 02C were selected in this study. Best band combination was defined for each satellite data, which based on union entropy; optimum index factor and correlation coefficient matrix after analysis each band entropy of three domestic high resolution satellite data. Five different fusion methods which are PC Spectral Sharpening（PC）, Color Normalized（Brovey）, Gram-Schmidt Spectral Sharpening（GS）, IHS transformand Multiplicative transform（MT） were applied in three kinds of domestic high resolution satellite images fusion. Furthermore, we analyzed band average gradient and image union entropy changes result before and after image fusion.The result show that optimum band combination of GF-1 areband4 for red, band3 for green, and band1 for blue. Band4 for red, band3 for green, and band2 for blue are the best bandcombination for ZY-3. The optimum band combination for ZY1 02C are band3 for red, band2 for green, and band1. The union entropy of color normalized fusion result was less than before image fusion in all three domestic satellite data, and the other four fusion result of union entropy were more than before fusion. According to these results, the best fusionmethod of GF-1, ZY-3 and ZY1 02C were Gram-schmidt spectral sharpening, PC spectral sharpeningand IHS transform, respectively.