融合细菌觅食趋化算子的生物地理学优化算法

摘要

A hybrid biogeography-based optimization ( HBBO) algorithm based on chemotaxis operator of bacterial foraging algorithm ( BFO) was proposed for solving problems in high dimensional multi-level col-or image thresholding. Firstly, some mutation operations were blended into the original migration operator to get a new migration operator. Secondly, a one-step length chemotaxis operator from BFO was combined with the new migration operator. In addition, a greedy selection method was used to substitute the original elitist selection approach to form a hybrid BBO. Finally, the HBBO algorithm was applied to high dimen-sional Tsallis entropy multilevel color image thresholding. Experimental results of image segmentation showed that the HBBO obtained more significant optimization performance, faster running speed and stronger stability compared with MABC, IDPSO, MBFO and BBO-M.%针对高维多阈值彩色图像分割中由于维数高带来阈值搜索困难等问题,提出了一种融合细菌觅食算法(BFO)趋化算子的混合生物地理学算法(hybrid biogeography-based optimization, HBBO)。首先构建一种嵌入变异操作的迁移算子,去掉BBO算法原有的变异算子；其次将细菌觅食算法中具有较强局部搜索能力的趋化算子的趋化步长固化为1,将此趋化算子与改进的迁移算子融合,并将精英保留策略换成贪婪选择算子,形成混合BBO算法；最后将HBBO算法应用到高维Tsallis熵多阈值彩色图像分割中。实验结果表明,与目前的MABC、IDPSO、MBFO和BBO-M算法相比,HBBO算法在高维多阈值图像分割中有更好的优化性能、更快的运行速度和更强的稳定性。