基于高分一号卫星影像的多种融合方法比较

Comparison of Fusion Methods Used for GF-1 Satellite Image

以扬州地区为试验区,利用高分一号(GF-1)卫星全色及多光谱影像数据开展融合试验,将PCA、GS、Pansharping 3种传统融合方法与Elbakary、Proxy-sharpening和SFIM等以全色波段为尺度因子的融合方法进行对比,从光谱继承性和空间融入度两方面分析融合效果,寻找适合GF-1号卫星影像的较佳融合方法。结果表明,传统融合方法在光谱继承性和空间融入度两方面的评价结果中保持一致,且Pansharping方法在定性和定量分析中都为最优;以全色波段为尺度因子的融合方法其空间融入度随着光谱继承性的降低而增加;基于不同移动窗口大小的SFIM方法其光谱继承度和空间融入度均属最优,其中,移动窗口为5×5和7×7时,整体融合精度高于其他方法。

Three traditional fusion methods including PCA, GS, and Pansharping were used to compare with the other three methods including Elbakary, Proxy-sharpening and SFIM whose scaling factor is panchromatic band based on the GF-1 satellite image data on January 14th,2014 in Yangzhou. Furthermore, two kinds of evaluation indexes were used to evaluate the fusion results qualitatively and quantitatively, including the spectral information inheritance indexes and the integration degree of spa- tial details indexes, so as to find the best method which is suitable for GF-1 satellite image. The results showed that the tradi- tional fusion methods kept consistency in two aspects of evaluation results and Pansharping method presented the optimum in qualitative and quantitative analysis. For the other three fusion methods scaling factor with panchromatic band showed the mu- tex in the two aspects, namely space into degree increased with the loss of the spectral inheritance. Based on the different moving window size, SFIM method can get the fusion results with the best spectral information inheritance and the most abundant spa- tial detail respectively, and on the condition of moving ＂ , window s size 5 × 5 and 7 × 7, the overall accuracy of this method is higher than the others,so that the SFIM method is a bettor fusion method for GF-1 satellite image.