压缩传感及其在医学图像融合中的应用  

Compressed sensing and its application in medical image fusion

作　　者：陈柘[1] 钟晓荣[1] 张晓博[1] 

机构地区：[1]长安大学信息工程学院,陕西西安710064

出　　处：《传感器与微系统》2013年第9期149-152,共4页Transducer and Microsystem Technologies

基　　金：中央高校基本科研业务费专项资助项目(CHD2010JC027)

摘　　要：为适应压缩传感成像技术的发展,降低融合运算对计算资源的需求,提出一种压缩传感域医学图像融合方法。算法利用双树复数小波变换具有的近似平移不变性、多方向的特性,以其作为稀疏分解基对图像做稀疏分解;分解后得到的系数,经由随机抽取哈达码块矩阵生成观测矢量;对得到的随机观测矢量,采用加权平均的方法进行融合;再经梯度投影重构生成融合图像的小波分解系数;最后,由逆双树复数小波变换生成融合后的图像。实验结果表明:所提算法可获得好的融合质量,并提高融合计算效率。In order to adapt to development of compressed sensing imaging technology,and reduce requirement of fusion arithmetic on computational resources,a medical image fusion algorithm in compressed sensing domain is proposed.Owing to the properties of dual tree complex wavelet transform（DT-CWT）,such as nearly translation invariance and multi-direction,the DT-CWT is used as sparse basis in image decomposition,then after decomposition,the coefficients are measured by scrambled block Hadamard matrix to produce observed vector,and the observed vectors are fused by weight averaging,then wavelet coefficients of fusion image are obtained by gradient project reconstruction.Finally,inverse dual tree complex wavelet transform is performed to produce fused image.Experimental results show that the proposed algorithm can obtain good fusion quality and at meanwhile improve efficiency of fusion computation.

关 键 词：医学图像融合 压缩传感 梯度投影 双树复数小波变换 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]