【摘要】 与其他进化算法相同,人工蜂群算法也会在搜索后期由于无法产生新位置而出现搜索停滞现象。基于此弱点,本文以两个食物源的中心位置为基准点,两者之间的方差为前进步长,提出一种基于分布更新的人工蜂群算法。此外,针对雇佣蜂和侦查蜂的不同特性,为其采用不同的食物源选择方式,使得算法既可以保证全局搜索,又可以加快收敛速度,标准测试函数上的实验结果验证了本文所提方法的有效性。最后,为解决传统灰度图像分割问题中由于暴力搜索所造成的耗时较长现象,本文以最大类间方差法(OTSU)作为评价准则,采用智能优化算法来寻找最优阈值。实验结果表明,本文所提出的改进人工蜂群算法不仅可以缩短计算时间,同时也取得了比其他进化算法更高的分割精度。更多还原

【Abstract】 Similar to other evolutionary algorithms,artificial bee colony cannot produce new positions at the later stage of optimization,which contributes to stagnation situation. To address this concerning issue,this paper utilizes two food sources’ central position to function as a basis point,and their deviation to determine the step size,and proposes a novel artificial bee colony algorithm with distribution-based update strategy. Besides,concerning the different characteristics of employed bees and onlooker bees,this paper also employs different food source selection mechanisms for these two types of bees to guarantee global search and accelerate convergence speed simultaneously. Experimental results on benchmarks demonstrate the proposed algorithm’s effectiveness.Finally,to handle the time-consuming phenomenon caused by exhaust search in traditional threshold-based image segmentation problem,this paper adopts intelligent optimization algorithms to search the optimal thresholds with OTSU function as the evaluation metric. Segmentation results on a set of images shows that the proposed artificial bee colony algorithm not only shortens the computation time but also achieves best segmentation accuracy when compared with other evolutionary algorithms.更多还原