多模态医学图像的融合研究  

THE STUDY OF THE MULTI-MODALITY MEDICAL IMAGE FUSION

作　　者：王修信[1] 张大力[2] 

机构地区：[1]广西师范大学物理与信息工程学院,广西桂林541004 [2]清华大学自动化系,北京100084

出　　处：《广西师范大学学报（自然科学版）》2004年第2期10-13,共4页Journal of Guangxi Normal University:Natural Science Edition

基　　金：广西教育厅科研基金资助项目

摘　　要：图像融合作为一种有效的信息融合的技术,已广泛用于医学图像、军事、遥感、机器视觉等领域.基于小波变换的图像融合是一种新的多尺度分解像素级融合方法,利用小波变换分别对CT,MRI医学图像进行分解处理,按照融合规则构造融合图像对应的各小波系数,再根据融合图像的各小波系数重构融合图像,重构后的融合图像完好地显示源图像各自的信息.实验图像使用互信息量化判据来评价融合效果,结果表明小波变换比传统的像素级加权平均融合算法效果更好.Being an available method of information fusion,image fusion has been used in many fields such as medical images,military applications,remote sensing and machine vision.The image fusion based on wavelet transform is a novel pixel-level image fusion scheme on multi-scale decomposition.The medical image CT and MRI were decomposed by means of the wavelet transform.Then the wavelet coefficients of the image fusion were formed with the fusion scheme.Finally the fused image was constructed with the wavelet coefficients.The fused image showed the preservations of the details of each input image successfully.The fused image was evaluated with the mutual information criterion quantitatively;the result indicated that the fusion scheme on wavelet transform was more effective than that generally on weighted mean.

关 键 词：医学图像 融合 小波变换 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]