非下采样Contourlet变换的图像融合及评价  

Image fusion based on nonsubsampled Contourlet transform and evaluation

作　　者：李其申[1] 李俊峰[1] 江泽涛[1] 

机构地区：[1]南昌航空大学计算机学院,南昌330063

出　　处：《计算机应用研究》2009年第3期1138-1141,1150,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(60673055)

摘　　要：分析和研究了非下采样Contourlet图像表示方法及其在图像变换中的优点,提出一种基于非下采样Contourlet变换的图像融合方法。首先将待融合源图像分解成不同尺度、多方向的频带;然后采取不同的融合方法对分解的高低频分量进行融合处理,低频系数采取局部能量优先的加权法融合,高频系数则采取局部梯度优先的加权法融合;最后将融合的各频带进行逆非下采样Contourlet变换得到融合图像。实验表明,在几种不同的客观评价标准下,该方法优于传统的小波域中的融合效果,能有效消除小波变换带来的光谱扭曲和假边缘现象。This paper introduced the characteristics of nonsubsampled Contourlet transform(NSCT),as well as its advantages in image transformation.NSCT was built upon nonsubsampled pyramids and nonsubsampled directional filter banks and it was a flexible multiscale,directional multiresolution and shift-unvariant image transform.Proposed a novel image fusion algorithm based on NSCT.Firstly,decomposed each source image in highpass subbands and lowpass subbands by NSCT.Secondly,used different fusion rules in different su...

关 键 词：图像融合 非下采样CONTOURLET变换 小波变换 光谱扭曲 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]