摘要
为了更准确地拟合图像的目标与背景的灰度级分布并分割出图像的目标部分,采用基于参数的阈值估计方法,提出了基于对数正态分布的粒子群EM混合算法,设计了对数正态分布参数的粒子群算法、EM算法和粒子群EM混合算法,给出了对数正态分布参数的计算过程。研究结果表明:对数正态混合分布能够很好地拟合一类图像的目标与背景的灰度级分布,粒子群EM混合算法具有较好的收敛性。该研究成果有助于解决一类图像的目标与背景的分割问题。
Based on the parameter-based threshold estimation method,this paper presents a particle swarm optimization and expectation maximization mixed algorithm based on the lognormal distributions in order to fit the gray-level distribution of object and background of an image accurately and to segment the image into object and background.Also the particle swarm optimization algorithm,the expectation maximization algorithm and the mixed method of swarm optimization and expectation maximization are designed to determine the parameters of the lognormal distribution.Study results show that the lognormal distributions can model the grey-level distribution of object and background of a kind of images.The particle swarm and expectation maximization mixed algorithm converges very well.The study results contribute to solve the problem in the object and background segmentation of a kind of images.
出处
《辽宁工程技术大学学报（自然科学版）》
CAS
北大核心
2010年第6期1106-1109,共4页
Journal of Liaoning Technical University (Natural Science)
基金
浙江省自然科学基金资助项目(Y6100029)
福建省教育厅A类基金资助项目(JA09187)
关键词
阈值
粒子群算法
EM算法
对数正态分布
threshold
particle swarm optimization
expectation maximization
lognormal distribution
