基于经验模式分解和互信息多模态图像配准[J].docVIP

基于经验模式分解和互信息的多模态图像配准 苑津莎1，赵振兵1，高　强1，丁　杰2，刘　蒙1 (1　华北电力大学电气与电子工程学院保定071003；2北京邮电大学信息与通信工程学院北京100876) 摘　要：基于互信息的配准方法是目前多模态图像配准研究中的热点。提出了一种基于经验模式分解和互信息的多模态图像配准方法。首先通过理论分析得出通过求解剩余图像之间的变换参数即可获得原始图像之间的变换参数，从而论证了二维经验模式分解BEMD)应用于多模态图像配准的可行性，然后给出了图像配准方法的实现步骤。典型多模态图像配准实验结果表明此方法与传统互信息法和基于小波分解结合互信息的方法相比，旋转角度估计误差可以降低1个数量级，缩放参数的估计误差也有很大降低。表明方法获得了更高的配准精度。TN911.73　　文献标识码：A　　国家标准学科分类代码：510.4050 Multi-modal image registration based on empirical mode decompositionand mutual information Yuan Jinsha1, Zhao Zhenbing1, Gao Qiang1, Ding Jie2, Liu Meng1 (1 School of Electrical and Electronic Engineering, North China Electric Power University, Baoding 071003, China;2 School of Information and Communication Engineering, Beijing University of Posts and Telecommunication, Beijing 100876, China) Abstract：Image registration based on mutual information has received much attention in multi-modality image registration at present. A method of multi-modality image registration based on remnant image and mutual information was proposed. Firstly, through theoretic analysis it is found that the transformation parameters between original images could be acquired by solving the transformation parameters between remnant images, which demonstrates the feasibility that bidimensional empirical mode decomposition (BEMD) could be successfully applied to multi-modality image registration. Then the implementation steps of the proposed method are given. Experiment results of typical multi-modality images demonstrate that the estimation error of rotation transformation parameter for this method can be reduced by one order in magnitude compared with traditional mutual information and wavelet composition method, which indicates that this method can obtain higher accuracy and better registration effect. Key words：empirical mode decomposition; remnant image; mutual information; image registration; multi-modal image 1　引　　言 图像配准是指依据一些相似性度量决定图像间的变换参数，使从不同传感器、不同视角、不同时间获取的同一场景的幅或多幅图像，变换到同一坐标系下，在像素层上得到最佳匹配的过程[1-2]。它是