Using mutual information as the similarity measurement and EM as the optimization algorithm, this paper proposes a new method to get the spatial transformation parameters for medical image registra-tion.During the experiments,the golden-section-point one-dimensional search method is adopted in the lo-cal search steps.The experimental results show that the MI and EM based registration algorithm can easily escape the local optimal solution with good accuracy and efficiency.%采用互信息来描述图像配准的效果,应用类电磁机制优化算法搜索配准所对应的空间变换参数,并借鉴POWELL优化算法,在局部搜索过程中采用了黄金分割点一维搜索方法进行搜索,实验结果表明提出的算法可以很好的脱离局部最优解,同时可以达到较好的配准精度,并具有较好的时效性。
展开▼