高分二号影像融合方法比较研究

Comparison of the image fusion methods for GF-2

影像融合能有效提高影像的利用效率,为探索GF-2遥感影像的最佳融合方法,利用HSV、Brovey、PC、NND、GS 5种融合方法对GF-2影像进行融合,并采用定性与定量评价法对融合质量进行评价。结果表明,HSV和Brovey融合法在空间细节上保留较好,但目视效果较差且仅能融合3个波段,光谱信息损失较大,NND在目视效果上最好,在可见光波段和空间细节的保真效果最好,PC与GS融合影像标准差与原始多光谱较为接近,在细节上GS比PC效果好。综合定性与定量评价,NND在GF-2号影像融合上效果最好,GS次之,Brovey融合质量整体最差。

The image fusion can effectively improve the utilization efficiency of remote sensing images.In order to explore the best fusion method of GF-2 remote sensing images,the paper uses the HSV,Brovey,PC,NND and GS to fuse GF-2 images,and then uses the visual and quantitative evaluation methods to evaluate the fusion quality.The result demonstrates that HSV and Brovey perform better in spectral fidelity,but the visual effect is poor and only three bands can be fused,lossing great spectral information.NND is the best algorithm for image information and spectral fidelity of GF-2.The fusion result of PC and GS is close to the original multi-spectrum,and GS has better effect than PC in detail.Based on the visual and quantitative evaluation results,NND is the best algorithm on the fusion of GF-2 satellite image,followed by GS,and Brovey is the worst falgorithm among the five methods.