摘要
采用了一种模拟退火思想的粒子群算法与最大类间方差法相结合的快速阈值分割法对图像进行分割。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,通过粒子间的协作来获得最优阈值。为了提高收敛速度,把模拟退火的思想应用在粒子群算法中,最后仿真结论表明,该方法在继承标准粒子群算法原理简单、易于实现、协同搜索等优点的同时,还避免了标准粒子群算法的收敛速度慢问题,有更强的寻优能力,得到理想的结果的同时计算量大大减少。权衡分割精度和计算效率两个方面,文中方法不失为一种实用有效的图像分割算法。
A rapid Otsu＇s method based on simulated annealing particle swarm optimization algorithm was used for image segmentation.Particle Swarm Optimization（PSO） algorithm was used to search threshold vectors.Each particle represents a feasible threshold vector.Thus,the optimal threshold could be acquired by the cooperation of particle swarm.To get better convergence,simulated annealing idea was applied in PSO algorithm.Simulation experiment results demonstrated that this method retains the uncomplicated principle,simple-operation and coevolutionary search of standard PSO,and also solves the slow convergence problem of standard PSO,could acquire ideal results with less computation.The algorithm is practical and effective in image segmentation considering segmentation accuracy and computing efficiency.
出处
《计算机技术与发展》
2010年第5期83-87,91,共6页
Computer Technology and Development
基金
国家自然科学基金项目(50407017)
安徽省自然科学基金重点项目(2006KJ019A
2007KJ052A)
关键词
模拟退火
粒子群算法
最大类间方差法
阈值分割
simulated annealing
particle swarm optimization
Otsu
threshold segmentation
