| EM与PSO算法在图像统计模型拟合中的应用 |
| |
| 引用本文： | 白彦辉. EM与PSO算法在图像统计模型拟合中的应用[J]. 科学技术与工程, 2013, 13(1): 236-239 |
| |
| 作者姓名： | 白彦辉 |
| |
| 作者单位： | 1. 赤峰学院计算机与信息工程学院,赤峰,024000
2. 江苏师范大学电气工程及自动化学院,徐州,221116 |
| |
| 基金项目： | 江苏省江苏师范大学2010年度自然科学基金（10XLR27） |
| |
| 摘 要： | 针对期望最大值算法(EM)对图像统计模型初始值敏感和容易陷入局部极值的弱点,结合粒子群优化算法(PSO)全局寻优的特点,提出一种有效解决此问题的EM-PSO混合算法.该算法将粒子分为最优种群和进化种群,分别用EM算法和PSO算法进行更新.然后选取最优粒子群作为EM算法的初始值.仿真结果表明,用EM-PSO算法拟合图像统计模型比用EM算法拟合图像统计模型更准确.
|
| 关 键 词： | 粒子群优化 期望最大值 图像统计模型 |
| 收稿时间： | 2012-08-21 |
| 修稿时间： | 2012-08-21 |
The Application of EM and PSO Algorithm to Fitting the Image Statistical Model |
| |
| bai yan hui. The Application of EM and PSO Algorithm to Fitting the Image Statistical Model[J]. Science Technology and Engineering, 2013, 13(1): 236-239 |
| |
| Authors: | bai yan hui |
| |
| Affiliation: | 2(College of Computer and Information Engineering,Chifeng University 1,Chifeng 024000,P.R.China;College of Electrical Engineering and Automation,Jiangsu Normal University 2,Xuzhou 221116,P.R.China) |
| |
| Abstract: | Aiming at the disadvantages of expectation maximization (EM) algorithm, in parameters estimation of statistical image model, which is sensitive to initialization and is easy to trap local maximum, combining the whole search characteristics of particle swarm optimization (PSO) algorithm, an hybrid EM-PSO algorithm is proposed. The proposed algorithm partitions the swarm into the history optimum swarm and the evolution swarm, which are, respectively, updated by EM algorithm and PSO algorithm, in each iteration, selecting the history optimum swarm for the EM algorithm initialization. Comparing EM-PSO algorithm with standard EM algorithm in fitting the statistical distribution of image, results show that the proposed outperforms EM algorithm. |
| |
| Keywords: | particle swarm optimization (PSO) expectation maximization(EM) parameter estimation |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《科学技术与工程》浏览原始摘要信息 |
|
点击此处可从《科学技术与工程》下载全文 |
|
