第二代Curvelet变换与像素能量特征对比度结合的医学图像算法融合算法  

Medical Image Fusion Method Base on the Combination of the Second Curvelet Transform and Pixel Energy Feature Contrast

作　　者：代茵[1,2] 王宇义 

机构地区：[1]东北大学信息科学学院,沈阳110819 [2]东北大学中荷生物医学与信息工程学院,沈阳110819

出　　处：《小型微型计算机系统》2015年第12期2760-2762,共3页Journal of Chinese Computer Systems

基　　金：中央高校基本科研业务费项目(N130319001)资助

摘　　要：针对传统的多尺度医学图像融合算法大多没有突出待融合图像的特征,在图像细节信息较丰富情况下,并没有突出图像边缘等较多信息,融合效果并不理想的特点,提出基于第二代Curvelet变换与像素能量特征对比度结合的医学图像融合算法.源图像经第二代Curvelet变换后,其低频部分采用基于窗口的融合算法,高频部分采用每个窗口的Curvelet能量系数取最大的融合算法,最后通过Curvelet逆变换获得融合图像.实验表明,本文提出的融合方法图像包含源图像的信息量大,对源图像的边缘结构保持最好,与源图像信息相关程度较高且差异性较少,实验结果远优与传统的图像融合方法 PA法(像素平均法)、拉氏(Laplace)法、DWT(小波变换)法.Traditional multi-scale medical image fusion cannot embody the image＇s features, especially;when the image has lots of de- tails, and some image＇s edge information may be lost. Therefore, this paperpresents a new medical image fusion algorithmbased on the combination of the second Curvelet transform and pixel energy feature contrast. After the second Curvelet transform, the low frequency of the image uses the image fusion method basing on window to process and the high frequency uses each window＇s Curvelettransform maximum coefficient. At last, fusion image is computed by inverse Curvelet transform. Experimentsresults showthe image processed with this method maintains lots of source image information and keep its edge＇s structure. Moreover, the processed image interrelates and resembles to the source image. The results of the experiments demonstrate this method better than when compared with the tradi- tional image fusion methods, such asPA（ Pixel Average）, Laplace and DWT.

关 键 词：医学图像融合 第二代CURVELET变换 像素特征对比度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]