基于GF影像的不同融合方法对城市水体光谱保真度影响

发布时间：2018-03-05 14:57

本文选题：遥感影像　切入点：城市水体　出处：《北京工业大学学报》2017年05期 　论文类型：期刊论文

【摘要】：为了评价融合后遥感影像上城市水体的光谱保真度,以GF1和GF2影像为数据源,应用经典的4种融合方法:Brovey、Gram-schmidt Pan Sharpening、NNDiffuse Pan Sharpening和主成分分析(principal component analysis,PCA),对GF1和GF2的全色影像和多光谱影像进行融合,并选取融合前后影像上的城市水体样本点,采用平均值、均方根误差(root mean square error,RMSE)、平均相对误差(mean relative error,MRE)来分析融合前后影像上城市水体波段偏差和波形保真度.结果表明:在波段偏差上,GF1和GF2影像均是采用NNDiffuse Pan Sharpening方法最好;在波形保真度上,GF1影像采用NNDiffuse Pan Sharpening方法最好,GF2影像采用Gram-schmidt Pan Sharpening方法最好.融合后的城市水体光谱保真度较好,可以进一步开展定量参数反演研究.
[Abstract]:In order to evaluate the spectral fidelity of urban water body on the fused remote sensing image, GF1 and GF2 images were used as data sources. Using four classical fusion methods: 1 Broveyn Gram Schmidt Pan sharing Pan Sharpening and principal component analysis (PCA), the panchromatic and multispectral images of GF1 and GF2 are fused, and the samples of urban water body before and after fusion are selected, and the average values are adopted. RMS root error mean square error (mean relative error mean relative error MREs) was used to analyze the waveband deviation and waveform fidelity of urban water before and after fusion. The results showed that NNDiffuse Pan Sharpening was the best method for both GF1 and GF2 images. On the waveform fidelity, NNDiffuse Pan Sharpening method is the best and Gram-schmidt Pan Sharpening method is the best for GF2 image. The fused urban water body has better spectral fidelity, so quantitative parameter inversion can be carried out further.
【作者单位】：河南理工大学测绘与国土信息工程学院;中国科学院遥感与数字地球研究所数字地球重点实验室;环境保护部卫星环境应用中心;
【基金】：高分辨率对地观测系统重大专项(41-Y20A31-9003-15/17) 中国科学院院遥感与数字地球研究所所长青年基金资助项目(Y6SJ2100CX)
【分类号】：TP79

【参考文献】