Contourlet四叉树系数方向相关性的遥感图像融合算法

Research of Remote Sensing Image Fusion Method Based on the Contourlet Coefficients' Correlativity

近年来,遥感图像融合技术作为遥感图像处理的重要分支,在资源调查、环境监测和区域分析等领域受到广泛关注.遥感图像融合技术可以融合不同传感器获得的不同图像,获得一幅信息完整、表达准确的融合图像.Contourlet变换以其优越的非线性逼近特性和良好的多分辨率、各向异性、冗余性等特点成为处理二维及多维信号奇异性的有利工具,并广泛应用到图像融合领域.结合Contourlet变换的特征,深入分析了Contourlet系数的关联特征,提出了一种基于Contourlet四叉树系数方向相关性的遥感图像融合方法.首先对不同传感器所采集的遥感图像进行Contourlet变换,获得不同尺度下的系数分布;然后根据各尺度之间的系数满足四叉树结构关系、树中各结点的方向相关性表现出一致的特征,提出了一种新的系数相关性融合规则,能够自适应地计算融合加权系数,进而获取融合系数;最后对融合系数进行Contourlet逆变换,得到遥感融合图像.相比于传统遥感图像融合方法,新算法获得的图像信息量更加丰富,纹理更加清晰,具有较强的实用性.

In recent years, as an important branch of remote sensing image processing technology, remote sensing image fusion technologies have been widely applied, especially in the fields of resource exploration, environmental monitoring, region analysis and so on. The techniques can fuse different images from different sensors to an image which has complete information and accurate expression. Contourlet transform is comprehensively concerned in the discipline of remote sensing image processing for its excellent characteristics such as non-linear approximation, multi-resolution, time- frequency localization, multi-directional and anisotropy. In this paper, combining the directional characteristics of Contourlet transform, we analyze the correlativity attribute and propose a novel image fusion algorithm for remote sensing images based on Contourlet coefficients＇ correlativity. Firstly, we separately perform Contourlet transform on the intensity component of multi-spectral remote sensing image obtained by IHS transform, and panchromatic remote sensing image. Secondly, we propose the fusion priciple of self-adaption calculating fuesd weighting coefficients. Finally, the target image is obtained by reverse Contourlet transform and reverse IHS transform. Compared with the traditional fusion methods, our algorithm can enhance the spatial resolution of target image. Meanwhile, it preserves the spectral information of multi-spectral image well.