摘要
为了快速有效地得到图像的最佳阈值,基于人工免疫系统中的克隆选择原理,提出一种新的混合遗传算法,并将其应用于基于最大类间方差法的图像阈值分割问题。该算法用克隆选择代替标准遗传算法中的概率选择,根据抗体-抗原的亲和度对种群中的优良个体有选择的克隆增殖,并利用抗体浓度调节机制来抑制高浓度抗体、促进低浓度抗体,以保持种群中个体的多样性。从而避免了遗传算法陷入局部最优解,出现早熟收敛现象。仿真实验结果表明,该算法对多类图像的良好分割效果和较强的实用能力。
To obtain the best image threshold fast and effectively, based on clonal selection principle of artificial immune system, a novel messy genetic algorithm is presented based on the method of maximum classes square error for image threshold segmentation problem. To prevent running into local classic solutions, emerging premature convergence, the probability selection in standard genetic algorithm is replaced by clonal selection, according to antibody-antigen affinity, the excellent individuals in population are cloned and multiplicated selectively, and chroma modulatory mechanism is used to restrain high chroma individuals, promote low chroma individuals and keep diversity of individuals in population. Emulational experimental results show that this method has good segmentation effects and strong practicability.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第5期1070-1072,共3页
Computer Engineering and Design
基金
国家科技支撑计划课题基金项目(2007BAD33B03)
国家民委2007年科研基金项目(07XBE04)
宁夏自然科学基金项目(NZ0693)
关键词
克隆选择
遗传算法
早熟收敛现象
阈值分割
最大类间方差法
clonal selection
genetic algorithm
premature convergence
threshold segmentation
method of maximum classes square error
