多/高光谱遥感图像的投影和小波融合算法

A Fusion Method of Hyperspectral and Multispectral Images Based on Projection and Wavelet Transformation

将高光谱图像与高空间分辨率图像融合后,由于融合图像空间分辨率提高,改变了混合像元内地物组分比例,像元光谱信息较原高光谱图像光谱信息会出现"失真"现象。针对这种情况,考虑混合像元内成分变化进行图像融合,首先利用投影方法模拟多光谱图像得到高光谱图像,并将模拟高光谱图像与原高光谱图像利用小波方法进行融合,融合图像不仅增强了空间信息,而且对光谱信息进行一定的修正,从而提高了环境异常探测等一系列应用的精度。利用hyperion图像和SPOT 5图像进行融合试验,融合图像能够识别出87.2%的目标区域。

After the fusion of hyperspectral and multispectral images, the pixels spectral informa- tion in fused image is distorted because of its improved spatial resolution and change in the ground object components. In the context of this situation, the composition changes of the mixed pixels should be considered for image fusion. First of all, the multispectral image is used to simulate a hyperspectral image using the method of projection. While in the second step wavelet transfor- mation （WT） is used to fuse the simulated and original hyperspectral images. The fused image can not only enhance the spatial information, but also correct the spectral information, and thus can increase the application accuracies such as the environmental abnormal detection. The hyperion image and the SPaT 5 image are chosen to do the experiment of fusion, 87.2 percent of the target areas can be distinguished when making use of the fused image.