高光谱分辨率的高光谱图像与高空间分辨率的全色图像融合可以综合两类图像的优势。将全色图像与高光谱部分波段分别融合, 再合成假彩色图像是融合的有效思路。引进多分辨率分析框架, 首先对源图像进行小波变换, 得到低频系数与高频系数; 然后从中提取平均梯度特征和边缘特征分别作为两类系数融合的依据; 最后将融合后的系数经过小波逆变换还原为各波段融合图像, 进行假彩色合成得到最终的彩色融合图。实验结果表明, 与 PCA、HIS等经典方法相比, 本文方法不仅能够在保证融合效率前提下有效保持光谱信息, 提高融合图像的空间分辨率, 融合图像的标准差、平均梯度、信息熵等各项指标也均优于其他融合方法。

Fusion of hyperspectral image with rich spectral information and panchromatic image with high spatial resolution can improve the quality of both two images effectively. It is a good way to choose proper bands of hyperspectral data to be fused with panchromatic image respectively, and the results are used for false color image composition for the fusion of two resources of images. In this paper, multi-resolution analysis frame is taken for fusion. Firstly, wavelet decomposition is carried out to resource images; multi features including average gradient and edge information are extracted from low frequency coefficients of two kinds of images. Then the low frequency coefficient of the fused image is obtained with regional average gradient weight method, while the high frequency coefficient is fused conducted by the edge feature of panchromatic image. Besides, the sigh of frequency coefficient is considered to avoid effect offset. At last, the fusion image of each band is obtained by performing inverse wavelet transforms, and after false color image composition, the fusion image is finally got. The experimental results indicate that the fused image obtained by this method can maintain both the spectral and spatial information effectively and efficiently. Compared with classical methods such as PCA and HIS, the fusion image using this method has the best performance in both subjective and objective evaluation.