摘要
为充分利用灰狼算法良好的搜索能力并进一步提升算法的收敛速度和计算精度,提出融合多种策略的灰狼算法,分割Renyi熵多阈值图像。文章提出非线性收敛因子,提高算法收敛性;采用自适应搜索策略,自动调整迭代初期和后期的头狼等级划分方式,对个体位置更新范围进行前期扩展和后期压缩,从而加快了寻优速度;应用柯西变异策略,帮助种群跳出局部极值;对比分析此算法与其他元启发式算法的多阈值分割方法。结果表明:所提算法在同样迭代次数下,得到最低的适应度值,且迭代曲线下降速度最快;当达到与次优算法相同的最小收敛值时,迭代次数平均可节省64.2%;在峰值信噪比和特征相似性分割指标上,得到至少一项最高分。
In order to make full use of the good search capability of the grey wolf algorithm and further improve the convergence speed and calculation accuracy of the algorithm,a grey wolf optimizer fused with multiple strategies(MSGWO)is proposed for Renyi entropy multi-threshold images segmentation.A nonlinear convergence factor is proposed to improve the convergence of the algorithm,and an adaptive search strategy is used to automatically adjust the way of head wolf rank division in the early and late iterations,and the range of individual position updates is pre-expanded and post-compressed to speed up the search for superiority.In addition,cauchy variation strategy is applied to help the population jump out of local extremes and to compare and analyze this algorithm with other metaheuristic algorithms for multi-threshold segmentation methods.The results show that the proposed algorithm can obtain the lowest fitness value under the same number of iterations,and when it reaches the same minimum fitness value as the sub-optimal algorithm,the number of iterations can be saved by an average of 64.2%.At least one highest score is obtained on the peak signal-to-noise ratio and feature similarity segmentation index.
作者
李斐
朱晓磊
LI Fei;ZHU Xiaolei(School of Information and Electrical Engineering,Shandong Jianzhu University,Jinan 250101,China;Shandong Key Laboratory of Intelligent Building Technology,Jinan 250101,China;Integrated Electronic Systems Lab Co.,Ltd.,Jinan 250100,China)
出处
《山东建筑大学学报》
2023年第4期39-46,共8页
Journal of Shandong Jianzhu University
基金
山东建筑大学博士科研基金项目(X19021Z0101)。
关键词
灰狼算法
自适应搜索
柯西变异
RENYI熵
多阈值分割
grey wolf algorithm
adaptive search
cauchy variation
Renyi entropy
multi-threshold segmentation
