为了使融合后的多光谱图像在显著提高空间分辨率的同时,尽可能多地保持原始多光谱特性,提出了一种基于Contourlet变换系数局部特征的选择性遥感图像融合方法。根据多光谱和全色图像融合过程中Contourlet变换后的低频和高频部分融合目的的不同,对得到的近似和各层各方向的细节分量分别运用窗口邻域移动模板逐一计算相应区域Contourlet系数阵的不同局部特征量,然后选择适当的准则,对图像的近似和细节分量分别应用不同的策略在Contourlet系数域内进行选择性融合,通过Contourlet和亮度-色调-饱和度(IHS)逆变换得到融合的高分辨率多光谱图像。采用Landsat TM多光谱和SPOT全色图像进行的融合实验结果表明:提出的算法在显著提高空间分辨率的同时,又能很好地保持原始图像的光谱特征,并优于传统的融合方法。 In order to keep the original multispectral properties as much as possible while improving the spatial resolution of the fused multispectral image, a remote sensing image fusion method based on the local features of the contourlet transform coefficients is proposed. According to the difference of the low frequency and high frequency part after the Contourlet transform in multi-spectral and panchromatic image fusion, the corresponding neighborhood Contourlet coefficient matrix Then, the appropriate criteria are selected to apply different strategies to the approximation and detail components of the image respectively, and then the images are selectively fused in the Contourlet coefficient domain. The fused images are obtained by Contourlet and IHS inverse transform High-resolution multispectral images. The fusion experiments using Landsat TM multispectral and SPOT panchromatic images show that the proposed algorithm can significantly improve the spatial resolution and preserve the spectral characteristics of the original image better than the traditional fusion method.