基于SAPSO优化三维Otsu方法的医学图像分割算法(英文)  

Three-dimensional Otsu’s method for medical image segmentation based on a simulated annealing particle swarm optimization algorithm

作　　者：白杨[1] 

机构地区：[1]温州大学电子信息系,浙江省温州市325035

出　　处：《中国组织工程研究与临床康复》2008年第22期4380-4384,共5页Journal of Clinical Rehabilitative Tissue Engineering Research

基　　金：浙江省自然科学基金资助项目(Y104592);浙江省教育厅科研项目(20041032)~~

摘　　要：医学图像具有内容丰富多样、特征丰富、多尺度等特性,因此对医学图像的分割比一般图像的难度更大。针对上述问题,提出了基于改进粒子群优化三维Otsu方法的医学图像分割算法。由于三维Otsu方法计算量大,采用粒子群优化算法来搜索阈值向量,每个粒子代表一个可行的阈值向量,通过粒子群之间的协作来获得最优阈值。由于粒子群优化算法容易陷入局部最优解的的缺点,提出了模拟退火的粒子群优化方法,使其能够快速准确得到整体最佳解,还能保持粒子群算法求解速度快的优点。最后通过仿真实验得出了结论表明,所提出的方法不仅能得到理想的结果,而且计算量大大减少。Medical image has rich content, various features, and multiple dimensions. Therefore, it is more difficult to segment medical image compared with general image. Aiming at this, a three-dimensional Otsu＇s method based on an improved particle swarm optimization （PSO） algorithm has been purposed for medical image segmentation. Three-dimensional Otsu＇s method requires much computation. PSO algorithm can be used to search threshold vectors. Each particle represents a feasible threshold vector. Thus, the optimal threshold can be acquired by the cooperation of particle swarm. Because the PSO algorithm easily sinks into local optimization, so a simulated annealing particle swarm optimization （SAPSO） algorithm has been purposed. The three-dimensional Otsu＇s method based on SAPSO can rapidly and exactly get the entire optimal results. Simulation experiment results demonstrated that this method could acquire ideal results with less computation.

关 键 词：粒子群算法 OTSU方法 模拟退火粒子群算法 图像分割 数字医学 

分 类 号：R411[医药卫生—临床医学]