高分一号影像融合及在水体信息提取中的应用

GF-1 Image Fusion and Its Application in Water Body Information Extraction

针对适用于GF-1卫星影像水体信息提取的影像融合方法展开研究,通过对高分影像的全色及多光谱数据进行融合实验,比较常用的5种影像融合方法,从主观定性评价和客观定量评价两方面分析,筛选出融合后影像质量较优的方法。采用较优融合方法进行水体信息提取,对比总体分类精度和Kappa系数,得出最适合GF-1影像的水体信息提取融合方法。结果表明,5种融合方法中Pansharp变换和GS变换融合质量较好。分别进行水体提取后,Pansharp变换的总体分类精度和Kappa系数的值均高于GS变换,故Pansharp变换更适用于GF-1卫星影像在水体信息提取中的应用。

In this paper, we studied the image fusion method for GF-1 satellite images water body information extraction. We used the fusion of the full-color and multi-spectral data of high-resolution images to compare the five commonly used image fusion methods, from subjective qualitative evaluation and objective quantification. Then, we analyzed from subjective qualitative evaluation and objective quantitative evaluation two aspects, selected the method of image quality after fusion, and used the better fusion method to extract the water body information. Finally, we compared the overall classification accuracy and Kappa coefficient, and obtained the water body information extraction and fusion method which was most suitable for GF-1 image. The results show that among the five fusion methods, the Pansharp transform and the GS transform have better fusion quality. After the water extraction, the Pansharp transform has higher overall classification accuracy and Kappa coefficient than GS transform, so the Pansharp transform is more suitable for the application of GF-1 satellite image in water body information extraction.