面向土地覆盖分类的MODIS影像融合研究

MODIS影像的多波段及其1、2波段的250 m中等分辨率为大区域中空间分辨率的土地覆盖制图提供了可能。为了有效利用MODIS影像的空间和光谱信息,使用SFIM、HPF和PCA变换等遥感影像融合方法,分别采用MODIS影像的波段1(b1)和波段2(b2)对3~7(b3~b7)波段进行融合,并就融合影像的光谱保真度和分类精度对6种不同融合结果进行评价。结果表明不同的融合结果得到的分类精度均有不同程度的提高；3种融合方法中使用b2的融合效果均优于b1；SFIM变换在光谱失真较小的情况下能够较大程度地提高分类精度。因此使用b2的SFIM变换可以用于提高MODIS土地覆盖图的空间分辨率和精度。

MODIS Image Fusion for Land Cover Classification

There are many spectral bands in MODIS images ,as well as the spatial resolution is 250 m for channels 1 and 2,which make it possible for large area medium resolution land cover mapping. In order to utilize the spatial and spectral information effectively, taking band 1(b_1) and/or band 2(b_2) as the high resolution images and using the remote sensing image fusion methods of SFIM, HPF and PCA etc to fuse band 3～7(b_3～b_7). The results of different fusion were evaluated by spectral fidelity and classification accuracy. It was found that all the 6 fusion results can enhance the land cover classification accuracy to different de-gree;The results of b_2 are better than b_1;SFIM transform can improve classification accuracy to a large ex-tent with relative smaller spectra distortion. Therefore the SFIM transform with b_2 can be used to improve the spatial resolution and accuracy of MODIS large areas land cover classification.