多光谱和全色图像融合适用的超复数主元加权方法

提出了一种适用于多光谱与全色图像融合的超复数主元加权方法.该方法首先对全色图像的每个像素值进行矢量化,然后对由RGB表示的多光谱图像和矢量化的全色图像分别用超复数矩阵进行表示,该表示方法考虑了由RGB表示多光谱图像的矢量性,从而避免了IHS和PCA融合方法由于忽视多光谱图像的矢量性而导致的色彩失真.通过对超复数矩阵表示的全色图像和多光谱图像分别进行超复数奇异值分解,分别获得了这两个超复数矩阵的超复数奇异值,并对得到的奇异值进行主元分析,提出了用最大特征值对应的特征向量作为权值进行加权图像融合的方法.分析和仿真的结果表明:提出的方法不存在人眼可见的光谱畸变.而用各种现有图像融合评估方法的评估结果也表明:提出的方法优于IHS、PCA和小波变换等融合方法.

Hypercomplex principle component weighted approach to multispectral and panchromatic images fusions

A hypercomplex principle component weighted approach to multispectral and panchromatic images fusions is presented.Each pixel of a panchromatic image is firstly transformed into a vector.Such a vectorial panchromatic image and the multispectral images represented by RGB are then described by hypercomplex matrices respectively.Since those hypercomplex matrices can take the vectorial property of RGB components into account,the proposed approach can avoid the color distortion of the fused image happening in the IHS and PCA fusion approaches.The hypercomplex singular value decomposition approach is applied to the hypercomplex matrices of the vectorial panchromatic and multispectral images.By means of the principle component analysis of the hypercomplex singular values of two matrices,a method for fusing two images is obtained by weighting the eigenvector corresponding to the largest eigenvalue.Both the analysis and simulation results show that there is not the distortion of visible spectrums in the image fused by the proposed method.The evaluation results from the various methods representing the state of the art ones also show that the fusion results of the presented approach are better than those of the IHS,PCA and wavelet-based methods.