基于小波的遥感影像特征信息融合模式研究

Wavelet Based Fusion Model of Feature Information in Remote Sensing Images

在数字图像小波多分辨率分析的理论基础上,采用小波变换方法对高分辨遥感图像的目标地物边缘进行信息增强,然后与多光谱遥感图像进行特征信息融合.在融合过程中,首先对多光谱图像中的R、G、B3个波段的图像进行小波分解,得到相应的低频图像,并对特征增强后的高分辨率图像进行小波分解,再将分解后的高频图像分别与低频图像进行融合,然后经RGB合成为彩色图像.该方法既改善了图像的清晰度和分辨率,同时也保留了原图像的光谱信息.最后通过融合实验验证了上述结论.

Fusion of feature information in remote sensing （RS） images is generally based upon spectral and texture features in object images. This paper presents a feature fusion approach to RS images after enhancement of edges in the remote sensing image with multi-resolution wavelet analysis. The image resolution is improved. Instead of the conventional HIS-wavelet fusion, fusion processes of the low frequent components of the RGB channels and the high frequency components, both obtained with wavelet decomposition, are performed separately, and a new fusion image is composed of the new RGB channels. In this way, not only resolution of the multi-spectral image is improved, but also color features, which are important for multi-spectral fusion, are retained. Effectiveness of the proposed method is verified with experimental results.