基于改进蜜蜂觅食算法的多阈值图像分割  

Multi-Threshold Image Segmentation Based on Improved Bee Foraging Algorithm

作　　者：郭宏志 吕征南 张志成[3] GUO Hongzhi;LYU Zhengnan;ZHANG Zhicheng(Institute of Urban Safety and Environmental Sciences,Beijing Academy of Science and Technology,Beijing 100054,China;China Electronics Technology Taiji Group Corporation Limited,Beijing 100083,China;School of Artificial Intelligence,Beijing University of Post and Telecommunications,Beijing 100876,China)

机构地区：[1]北京市科学技术研究院城市安全与环境科学研究所,北京100054 [2]中电太极(集团)有限公司,北京100083 [3]北京邮电大学人工智能学院,北京100876

出　　处：《测控技术》2024年第9期28-34,共7页Measurement & Control Technology

基　　金：国家重点研发计划资助(2022YFF1302700)。

摘　　要：为解决传统多阈值图像分割方法搜索最优阈值效率低的问题,利用蜜蜂觅食算法中蜂群对蜜源的智能搜索机制,提高多阈值图像分割的最优阈值搜索效率。针对蜜蜂觅食算法邻域收缩过快易陷入局部最优问题,引入自适应邻域收缩策略,根据搜索陷入停滞的状态,自适应调整邻域收缩率,在保证算法搜索效率的同时,进一步提高算法寻优精度。仿真实验结果证明该方法提升了蜜蜂觅食算法的优化性能和算法鲁棒性,基于最大类间方差的图像多阈值分割实验验证了该方法的有效性。The intelligent searching mechanism of bee colony for honey sources in the bee foraging algorithm is utilized to promote the optimal threshold searching efficiency of multi-threshold image segmentation,which aimed to solve the problem of low optimal threshold search efficiency for the traditional multi-threshold image segmentation methods.An adaptive neighbourhood shrinking strategy is introduced for overcoming the bee fora-ging algorithm falling into local optimum caused by a rapidly shrinking neighbourhood,where the neighbour-hood shrinking rate can be adjusted dynamically according to the stagnation state of the searching process.The strategy ensures the searching efficiency of the algorithm and further improves the optimization accuracy of the algorithm.The simulation experimental results show that the proposed method improves the optimization per-formance and robustness of the bee foraging algorithm.The experiment of multi-threshold image segmentation based on maximum inter-class variance method verifies the effectiveness of the proposed method.

关 键 词：蜜蜂觅食算法 多阈值 图像分割 自适应邻域收缩 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP18[自动化与计算机技术—计算机科学与技术]