保持光谱信息的遥感图像融合方法研究

常用的遥感图像融合方法,如IHS变换法、Brovey变换法和主成分变换法等在实施图像融合时,均会有不同程度的光谱扭曲现象.探讨能有效保持光谱信息的EECN融合法.EECN融合法采用比值变换法,同时对参与融合的全色波段进行增强边缘,融合后的图像在光谱保持性能、分类精度等方面均较优.

A Spectral Preservation Fusion Technique for Remote Sensing Images

In fusion of a lower spatial resolution multispectral satellite images with a higher spatialresolution panchromatic one is quite valuable to identify objects. A variety of fusion approaches areavailable, such as the IHS(Intensity-Hue-Saturation)transform, the Brovey transform and theprincipal component analysis transform. However, these existing techniques cannot avoid colordistortion. In this paper, the principle of the Brovey transform fusion was reviewed. A spectralpreservation image fusion technique, called edge enhancement color normalized (EECN) was de-veloped to merge effectively lower spatial resolution multispectral images as well as a higher spatialresolution panchromatic one. In addition, a gray variance index (GVI) model was applied to eval-uate the capability of preserving spectral property for the fused images. The results of the EECNfusion image were convincingly demonstrated by an experiment using Landsat-5 TM and SPOTPanchromatic images of Fuzhou in Jiangxi, China. The visual evaluation and statistical analysiscompared between the EECN image and the Brovey image confirmed that the EECN images rep-resent almost the same color as the original multispectral images and the same space detail as theoriginal panchromatic images, the EECN technique is more powerful on preserving space detail,and reducing significantly the color distortion of the fused images which usually occurred when us-ing the Brovey approach.