基于改进飞蛾扑火算法的多阈值彩色图像分割

针对彩色图像多阈值分割存在计算量大、运行时间长等问题,在飞蛾扑火算法(Moth-Flame Optimization,MFO)的基础上,引入莱维飞行策略和自适应权重变化策略,提出LSMFO算法(Levy Self-adaptive Moth Flame Optimization)对最佳分割阈值进行优化搜索。为了验证该算法的有效性,选取4幅伯克利大学经典图像,将LSMFO算法与另外5种元启发式算法进行对比。应用Otsu方法进行多阈值图像分割实验,并用SSIM、PSNR、EPI三个指标评估分割后的图像效果。实验结果显示,LSMFO算法在指标衡量比较上整体水平优于其他算法,表明该算法运行时间短、分割精度高,能够有效解决彩色图像多阈值分割问题。

MULTI-THRESHOLD COLOR IMAGE SEGMENTATION BASED ON MODIFIED MOTH FLAME OPTIMIZATION ALGORITHM

There are some problems in multi-threshold segmentation of color images,such as large amount of computation and long running time.To solve this problem,we introduce Levy flight strategy and adaptive weight strategy based on moth-flame optimization(MFO),and propose LSMFO algorithm(Levy Self-adaptive Moth Flame Optimization)to search the optimal segmentation threshold.Four classic Berkeley images were selected to verify the effectiveness of the proposed algorithm.We compared LSMFO algorithm with five other meta-heuristics,adopted the Otsu method to perform multi-threshold image segmentation experiments,and applied SSIM,PSNR and EPI to evaluate the segmentation effects.The experimental results show that LSMFO is superior to other algorithms in the comparison of the whole of indicators and the proposed algorithm has the characteristics of short running time and high segmentation precision,which can effectively solve the problem of color images threshold segmentation problem.