基于双正交多小波变换的多光谱与全色图像融合研究

Fusion of Multispectral and Panchromatic Images Based on Biorthogonal Multiwavelet Transform

针对多光谱与全色图像融合中存在的光谱扭曲问题,提出了一种利用双正交多小波进行多分辨率分析,并结合平均与选择法处理小波高频系数的融合算法。该算法首先对已配准的多光谱图像进行IHS变换,然后分别对变换得到的强度分量I与全色图像进行双正交多小波分解,为增强融合图像的空间信息,对分解得到的高频系数利用平均与选择相结合的方法来确定,低频系数则通过邻域方差准则得到。最后由新的小波低频和高频系数重构并进行IHS逆变换得到融合图像。实验结果表明,该方法可以有效减少光谱扭曲,并提高图像的空间分辨率,保留图像中的边缘细节。

In order to solve the problem of spectral distortion existed in the panchromatic and muhispectral images fusion algorithm, a novel algorithm based on the biorthogonal muhiwavelet transform and the method of combination of average and selection is presented. First it decomposes the intensity components which obtained by IHS transform of the registered multispectral image and biorthogonal multi-wavelet transform of the panchromatic image. Then the method of combination of average and selection and the local variance rule is separately adopted to obtain new high frequency and low frequency coefficients to enhance the edge information of fused images. Finally these images are reconstructed with composite wavelet coefficients and by performing the inverse IHS transform. The experimental results show that this algorithm can improve the spatial resolution of multispectral image while reducing spectral distortion and still maintaining its edge detail.