基于动态自适应蚁群算法的MRI图像分割  

Segmentation of MRI Based on Dynamic and Adaptive Ant Colony Algorithm

作　　者：白杨[1] 孙跃[1] 王君[1] 周文俊[1] 胡宁萍 

机构地区：[1]温州大学城市学院,温州325000 [2]温州附二医,温州325010

出　　处：《计算机科学》2008年第2期226-229,共4页Computer Science

基　　金：浙江省自然科学基金资助项目(No.Y104592)准确测量脑部纵向变化的四维图像一致性分割算法;浙江省教育厅资助项目(20041032)

摘　　要：MRI图像分割在医学图像分析中具有极其重要的理论和应用价值。蚁群算法是一种具有离散性、并行性、鲁棒性和模糊聚类能力的进化方法。对目标边界模糊、目标灰度不均匀及目标不连续等情况的图像(如医学图像)分割,蚁群算法是一个比较好的选择。本文针对基本蚁群算法容易出现早熟和停滞现象的特性,提出了一种动态自适应蚁群算法,通过自适应的初始聚类中心调整策略和动态更新局部信息素浓度,使其收敛性和稳定性有一定的提高。实验证明改进的蚁群算法能够有效地分割MRI图像。Segmentation of MRI is very important in medical image analysis. Ant colony algorithm（ACA） is a kind of discrete,parallel and robust evolutionary algorithm with fuzzy clustering ability. To segment targets with blurry edges, intensity non-uniformity and discontinuity （such as medical images）, ACA approach is a good choice. A dynamic and a- daptive ant colony algorithm is presented in accordance with the defect of early variety and stagnation. The contribution of the algorithm includes an adaptive stratety of primary clustering center and a local updating for pheromone dynamically. Using this method can segment image fast and accurately. Experimental results show that the improved ACA is an effective MRI segmentation

关 键 词：蚁群算法 磁共振图像 图像分割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TU457[自动化与计算机技术—计算机科学与技术]