改进?鱼优化算法和熵测度的图像多阈值分割  

An improved remora optimization algorithm for multilevel thresholding image segmentation using an entropy measure

作　　者：刘庆鑫 李霓[2,3] 贾鹤鸣 齐琦 LIU Qingxin;LI Ni;JIA Heming;QI Qi(School of Computer Science and Technology,Hainan University,Haikou 570228,China;School of Mathematics and Statistics,Hainan Normal University,Haikou 571158,China;Key Laboratory of Data Science and Intelligence Education of Ministry of Edu-cation,Hainan Normal University,Haikou 571158,China;School of Information Engineering,Sanming University,Sanming 365004,China)

机构地区：[1]海南大学计算机科学与技术学院,海南海口570228 [2]海南师范大学数学与统计学院,海南海口571158 [3]海南师范大学数据科学与智慧教育教育部重点实验室,海南海口571158 [4]三明学院信息工程学院,福建三明365004

出　　处：《智能系统学报》2024年第2期381-391,共11页CAAI Transactions on Intelligent Systems

基　　金：国家自然科学基金项目(11861030);海南省自然科学基金项目(621RC511,2019RC176);海南省研究生创新科研课题(Qhys2021-190)。

摘　　要：针对传统图像多阈值分割方法存在效率低、分割质量差等问题,提出一种改进?鱼优化算法并结合熵测度(weight lens remora optimization algorithm,WLROA)的图像多阈值分割方法。针对?鱼优化算法易陷入局部极值等缺陷,引入透镜成像反向学习策略,生成透镜反向解来增加种群多样性,进而提高算法跳出局部极值能力;提出一种自适应权重因子,对个体位置进行自适应扰动,提高算法探索能力。以最小化交叉熵作为优化目标,利用WLROA确定最小交叉熵并获得相应分割阈值。选取部分伯克利大学分割数据集图像和遥感图像测试提出算法的分割性能,测试结果表明,WLROA与其他知名算法相比具有更好的分割效果,能够有效实现复杂图像的精确处理。To improve the poor segmentation quality of traditional image thresholding segmentation techniques,this study proposes an image multilevel thresholding segmentation method.This method is based on an improved remora optimization algorithm and entropy measure,specifically called the weight lens remora optimization algorithm(WLROA).First,lens opposition-based learning was used to generate the lens opposite solution.This approach bolstered population diversity and improved the algorithm’s ability to overcome local optimal solutions.Furthermore,an adaptive weight factor was introduced to perturb the individuals’positions appropriately.This modification aimed to improve the algorithm’s exploratory ability.The optimization objective was to minimize cross entropy.To achieve this,WLROA was used to determine the minimum cross entropy and obtain the corresponding thresholds.A selection of images from the Berkeley segmentation data set and remote sensing images were selected to assess the segmentation performance of the proposed algorithm.These results were then compared with those from other methods.The results revealed that,in comparison with other well-known algorithms,WLROA yielded better segmentation results and proved effective in accurately processing complex images.

关 键 词：图像处理 多阈值分割 ?鱼优化算法 最小交叉熵 透镜成像反向学习 自适应权重因子 全局优化 遥感图像 

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]