【摘要】 为了实现医学图像的快速准确配准,提出了基于群搜索算法,以及群搜索算法与Powell法相结合的医学图像配准算法,并以互信息作为图像配准的相似性测度,使用群搜索算法、群搜索与Powell法相结合的方法在指定的搜索空间求解配准所需的空间变换参数,从而实现图像的配准.实验结果表明:群搜索算法能够成功实现医学图像配准,与粒子群算法相比,群搜索算法用于医学图像配准能够得到更高的配准精度;群搜索算法与Powell法结合,能够在减少目标函数计算次数的基础上,极大地提高图像配准的精度.更多还原

【Abstract】 In order to implement medical image registration both fast and precisely,the algorithms for medical image registration using group search optimizer( GSO) and group search optimizer combined with Powell’s method are proposed. With mutual information employed as similarity measure of the images,GSO or GSO combined with Powell’s method are then introduced to solve the space transformation parameters for registration in a specified search space,and therefore,two images can be registered. Experiment results show that GSO is proper for medical image registration,and compared with particle swarm optimizer( PSO),GSO receives a better effect;GSO combined with Powell’s method can enhance the precision of registration markedly while reducing the times of calculation of the objective function.更多还原