摘要
在标准粒子群优化算法的每一次迭代中,粒子都是同时更新,然而在现实世界中(比如鸟群)粒子的更新并不是同时的.模拟现实的鸟群更新,找到一种异步粒子群优化算法.异步粒子群优化算法是将粒子的更新顺序进行改进,在每次迭代中将所有粒子按适应度的优劣排序,从而每个粒子在更新时都能利用到当代群体的信息,所以算法更易于收敛.提出一种基于异步粒子群优化算法的图像分割方法,用异步粒子群优化算法自适应选取图像的分割阈值.实验表明,与基本的粒子群优化算法相比,该算法比较稳定,易于收敛到最优解,分割速度较快.
In a standard PSO algorithm all particles will be updated at the same time during the course of iteration, while in real-world swarm (such as flock of birds) not all particles are updated in the same time. A novel particle swarm opti- mization algorithm is proposed by simulating the real-world swarm. A asynchronous particle swarm optimization is based on improving the update order of particles, which sorts all particles lay computing the fitness, so particles update those positions and velocities continuously by making use of currently available information. A novel image segmentation method based on asynchronous particle swarm optimization is provided, which finds the threshold by asynchronous particle swarm optimizatiort Numerical experiments show that the asynchronous particle swarm optimization per/otto.s similarly and sometime even better than standard approaches for the considered problems.
出处
《微电子学与计算机》
CSCD
北大核心
2009年第4期174-177,共4页
Microelectronics & Computer
基金
江苏省高校自然科学基础研究项目(08KJB520003)
浙江大学CAD&CG国家重点实验室开放课题基金项目(A0704)
关键词
图像分割
粒子群优化算法
异步
更新顺序
image segmentation
particle swarm optimization
asynchronous
update order