基于非下采样Contourlets的CT/MRI图像自适应融合  

Adaptive CT/MRI Image Fusion Based on Nonsubsampled Contourlets

作　　者：杨晓慧[1] 朱秀阁[2] 

机构地区：[1]河南大学数学与信息科学学院应用数学研究所,河南开封475004 [2]河南大学计算机中心,河南开封475001

出　　处：《计算机技术与发展》2008年第12期116-119,共4页Computer Technology and Development

基　　金：河南省创新型科技人才队伍建设工程基金(084100510012);河南省自然科学基金(0611053200);河南省教育厅自然科学基金(2008B510001)

摘　　要：结合人类视觉特性,针对CT/MRI医学图像的特点,提出了一种基于非下采样Contourlet变换的图像融合算法。先对源图像作非下采样Contourlet变换,完成图像的多尺度分析和方向分析。充分考虑各尺度分解层的系数特征,对低通子带,基于评价准则最优,采用免疫克隆选择优化策略迭代获取近似最优融合权值;对高通子带则选取绝对值最大作融合。实验结果表明:分别与基于小波、非下采样小波,以及Contourlet的融合结果相比较,文中融合算法获得的融合图像边缘的清晰度,以及整体的对比度都有所改善。A novel adaptive image fusion scheme hazed on nonsubsampled Contourlet transform （NSCT） was presented, which combines with human visual system and is designed for CT/MRI medical images. Firstly, the NSCT is used to perform a multi - resolution analysis and directional analysis. By taking advantage of the characteristics of the coefficients on different scales, immune clonal selection strategy was introduced in low- pass sub - bands to obtain asymptotic optimal fused weights adaptively. To high - pass sub ：- banda, select the maximal absolute values. Comparisons are made with wavelets- based, nonsubsampled wavelets- based and Contourlets - based fusion methods, respectively. Experimental results show that the fused images based on the fusion technique have improvements on definition of edges and contrast of whole image.

关 键 词：图像融合 非下采样Contouflet变换 CT/MRI图像 

分 类 号：TN911.13[电子电信—通信与信息系统]