基于支持向量值轮廓波变换的遥感影像融合

Remote Sensing Image Fusion Based on SupportVector Value Contourlet Transform

本文研究基于支持向量值轮廓波变换的遥感影像融合.首先将支持向量值滤波器和方向滤波器组进行合理搭配,构造支持向量值轮廓波变换,该变换具有平移不变、泛化能力好、捕捉奇异性能强等特性.然后利用该变换对多源遥感影像进行多尺度、多方向、多分辨率分解,在不同的分解水平上利用基于区域能量和基于区域的轮廓波对比度方法进行融合.实验结果表明,利用该变换进行遥感影像融合时,能在不降低空间分辨率的情况下有效保留源影像的光谱信息.

Remote sensing image fusion based on support vector value contourlet transform is researched in this paper.Firstly support vector value filter and directional filter bank are rationally matched to construct support vector value contourlet transform,which has advantages of shift invariant、good generalization and strongly catching singularity ability.Then the transform is used to decompose multi-source remote sensing images at multi-scale、multi-direction and multi-resolution.At different levels,images are fused based on regional energy and regional contourlet contrast.Experiments show that,when remote sensing images being fused by using this transform,spectrum is effectively preserved while space resolution being not reduced.