比值法保全多光谱图像颜色的研究

A Ratio Algorithm for Retaining Color of Multi-spectral Image

针对遥感图像融合Brovey变换法存在颜色失真的现象,提出了一种低通比值融合法,该融合法首先对高几何分辨率的全色波段进行低通滤波,尔后将低分辨率多光谱图像与全色波段图像相乘,再除以滤波后的全色波段图像,便得到融合图像。从辐照的角度证明了该低通比值融合法具备理论基础,并从目视评价、定量分析、分类精度证实了该低通比值融合法优于Brovey变换法,该低通比值融合法是一种能较好地保全低分辨率多光谱图像颜色的融合方法。

Remote sensing technology is often applied to fleetly investigate natural resources and environment; the fused production might acquire detailed resources and environment content from multi-source remote sensing data in less cost by the means of merging low resolution multi-spectral data with high resolution panchromatic data, and obtain the synthetical information derived from both of the multi-spectral and panchromatic data. These conventional fusion methods including IHS transform, Brovey transform and principal components transform could merger two optical image data of different resolutions--a high spatial resolution panchromatic image and a low spatial resolution but multi-spectral image.But their production＇s colors differ from that of the low spatial resolution multi-spectral image. This paper analyzed the Brovey transform, and then proposed a improved method called panchromatic image low-pass filter based ratio fusion （Low-pass RF）, that production＇s color is accordant with that of the multi-spectral image. This paper discussed its theory mechanism at terms of radiancy. The results of visual interpretation and quantitative analysis to gray variance index show that the Low-pass RF transform have a capability of retaining color from multi-spectral image. Classification accuracy of the fused image by Low-pass RF transform is 82.45%, more 5.69% higher than the Brovey image. Therefore, Low-pass RF transform is more excellent than Brovey transform.