采用双树小波变换的医学图像融合方法及实现  

ON MEDICAL IMAGE FUSION METHOD USING DUAL-TREE WAVELET TRANSFORM AND ITS IMPLEMENTATION

作　　者：李洪海[1] 朱霞[1] 

机构地区：[1]淮阴工学院电子与电气工程学院,江苏淮安223003

出　　处：《计算机应用与软件》2012年第11期292-294,327,共4页Computer Applications and Software

基　　金：淮安市科技计划项目(SN1045);淮安市科技局项目(HAG09052)

摘　　要：双树小波变换具有平移不变性及良好的方向性,将其引入到医学图像融合能够较好地提取原始图像的特征,为融合图像提供更多的信息。提出基于双树小波变换的医学图像融合算法,该算法对图像进行多尺度和多方向的分解,在相应尺度上利用融合算法融合,最后进行重构得到融合结果。通过实验仿真及客观的图像融合评估准则,实验结果表明了双树小波分解能够得到比树状小波分解更好的融合效果,信噪比提高了很多,均方差明显减少,充分保留了图像中的细节信息,减少了融合的复杂度。Dual-tree wavelet transform has translation invariability and good directionality.When it is introduced to medical image fusion,the characteristics of original images are extracted better and more information for fusion is obtained.An algorithm for medical image fusion is proposed based on dual-tree wavelet transform.Firstly,the source image is decomposed in multi-scale and multi-direction,then fused with fusion algorithm on corresponding scales,and finally reconstructed to obtain the fusion results.Experimental simulation and objective image fusion assessment criteria are employed.The experimental results show that the dual-tree wavelet decomposition can achieve better fusion effect than the tree-structure wavelet decomposition,the signal-to-noise ratio has improved a lot,the root mean square error has been significantly reduced,plenty of detailed information of the image is retained,and the complexity of fusion is reduced as well.

关 键 词：图像融合 小波变换 树状小波 双树小波 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]