摘要
图像配准一直是图像研究领域的热点话题,互信息的配准方法由于其精度高、鲁棒性强等特点,成为图像配准中的常用方法。但其目标函数存在局部极值问题。针对这个问题,提出一种量子行为的粒子群优化算法(QPSO)和Powell法相结合的多分辨率搜索优化算法。QPSO参数个数少,其每一个迭代步的取样空间能覆盖整个解空间,能保证算法的全局收敛,因此可以有效地解决Powell算法的缺点。该算法将量子行为的粒子群优化算法(QPSO)与Powell法结合起来对二维的MRI图像进行配准。实验结果表明,该方法能够有效地克服互信息函数的局部极值问题,并提高了配准精度和速度。
Image registration has always been the hot topic in image research field, the mutual information registration method becomes a commonly used method in image registration because of its high precision and good robustness. Unfortunately, its objective function has the problem of local extremum. Aiming at this issue, this paper proposes a multi-resolution search optimisation algorithm which combines the QPSO with Powell＇ s method. Since QPSO has less parameter, and the sampling space Of its every iterative step can cover the whole solution space and can guarantee the global convergence of the algorithm, thus the QPSO can effectively overcome the drawback of Powell＇ s method. The proposed algorithm registers the 2D MRI image by combining the QPSO algorithm with Powell＇ s method. Experimental results show that this method can effectively solve the local extremum problem of mutual information function, and effectively improves the accuracy and speed of registration as well.
出处
《计算机应用与软件》
CSCD
北大核心
2014年第7期237-240,共4页
Computer Applications and Software
