基于互信息医学刚性图像配准方法的研究

摘要

目的:基于互信息的配准方法是医学图像配准领域的重要方法,具有鲁棒性,精度高等优点.本文探究医学刚性图像配准的有效算法和关键技术.方法:基于互信息配准方法,利用Powell多参数算法和改进的PV插值算法,得到两幅图像之间的最大互信息和最佳配准参数.结果:二维磨牙CT图像配准实验表明,配准速度快,精度提高,验证了插值方法的有效性.结论:方法和算法可提高配准速度,能有效抑制互信息目标函数的局部极值.%Objective: Registration method based on mutual information is an important method in the field of medical image registration, which is accepted as the most robust and accurate. To explore the effective algorithm and the key technology of medical rigid image registration in this paper. Methods: The registration approach was based on mutual information. We used the multi-parameters Powell algorithm and the improved Partial Volume Distribution interpolation algorithm. The maximum of mutual information and the best registration parameters of between the two images were obtained. Results: A registration experiment for two dimensional CT image of the jawtooth shows that the registration speed is faster and precision is improved. The interpolation method has been verified. Conclusions: The method and algorithm can improve the speed of the registration and avoid the local extremum problem of objective function of mutual information effectively.