基于多特征的遥感图像融合算法

摘要

The remote sensing image fusion algorithm based on multiscale transform can not extract details from source images effectively; so a new remote sensing image fusion algorithm based on non⁃subsampled contourlet transform ( NSCT) domain is proposed. Firstly, the multi⁃spectral image was transformed into HSI ( Hue⁃Saturation⁃Intensity) color space, and the NSCT transform was employed to decompose the Intensity component and panchro⁃matic image into multiresolution representation;Secondly, the fusion rule of selecting maximum absolute pixel val⁃ues was used for the low frequency sub⁃band coefficients, while for the high frequency subband coefficients, the multi⁃feature fusion rule was designed; the fused image was reconstructed by inverse NSCT transform and inverse HSI transform. Experiments and their analysis show preliminarily that the fusion method proposed can improve spa⁃tial resolution and keep spectral information simultaneously and that there are improvements both in visual effects and quantitative analysis compared with the traditional HSI tansform method, the contourlet transform based fusion method, and the NSCT transform based fusion method.%针对基于多尺度几何变换的遥感图像融合算法细节表现能力不足的缺陷，提出了一种新的基于多特征的遥感图像融合算法。首先，对多光谱图像进行HSI变换，将得到的亮度分量和全色图像分别进行非下采样的Contourlet变换（ NSCT），得到低频和高频子带系数；然后，对低频子带系数采用像素绝对值选大的规则进行融合，对于高频子带系数的选择，考虑到不同的因素如（方差、能量、平均梯度）对图像质量的影响不同，提出了一种基于多特征的融合规则；最后，对融合后的低频和高频系数分别进行了逆NSCT变换和逆HSI变换得到融合图像。实验结果证明，该方法可以有效将全色图像的空间信息注入到多光谱图像中，并与HSI变换、Contourlet变换等融合算法相比，该方法在主观和客观评价上优于其他几种融合方法，具有更好的融合效果。