Abstract: Aiming at the problem that spatial information and spectral information are not well compatible in remote sensing image fusion, a remote sensing image fusion method is proposed based on wavelet packet transform with Sobel operator. In this method, multi-spectral image and high resolution image are decomposed by wavelet packet transform. According to the threshold, it uses different fusion rules to get low-frequency coefficient of wavelet, uses Sobel operator to extract high-frequency characteristic value of image, and adopts the best value method to get high-frequency coefficients. Experimental results show that the proposed algorithm is superior to the traditional IHS（Intensity, Hue, Saturation） transform, wavelet transform and a combination of both methods. This method not only better retains the spectral information of image, but also enhances the image details, edge feature, and clarity of the image has been improved very well.

Key words: wavelet packet transform, Intensity, Hue, Saturation（IHS） transform, Sobel operator, spectral preservation

摘要： 针对遥感图像融合时图像的空间信息与光谱信息不易兼容的问题，在小波包变换的基础上提出了一种基于Sobel算子的图像融合算法。该方法将多光谱图像与高分辨率图像进行小波包变换，根据阈值选用不同的融合准则得到小波低频系数，利用Sobel算子提取图像高频特征值，采用最值法获取高频系数。实验结果表明，所提算法优于传统的HIS（Intensity，Hue，Saturation）变换、小波变换以及两者结合的方法，在较好地保留图像光谱信息的同时，进一步增强图像的细节信息、边缘特征，从而提高图像的清晰度。

关键词: 小波包变换, 亮度或强度、色调、饱和度（IHS）变换, Sobel算子, 光谱保持