摘要
针对传统蝙蝠算法在图像阈值分割时存在的分割精度低、效果差、算法收敛速度慢等缺点,提出基于分数阶混合蝙蝠算法的Otsu图像分割算法。将分数阶微积分、天牛须搜索与传统蝙蝠算法相结合,利用分数阶微积分的遗传特性均衡蝙蝠的全局搜索过程,提高寻优能力。利用天牛须搜索丰富蝙蝠局部搜索时种群的多样性,加快高精度下算法的收敛速度。将二维Otsu算法的离散度测度矩阵作为寻优的目标函数进行图像阈值分割。实验结果表明,改进后的算法提升了图像的分割精度与效果,且有着更快的收敛速度和良好的健壮性。
Aiming at the disadvantages of traditional bat algorithm in image threshold segmentation,such as low segmentation accuracy,poor effect,slow algorithm convergence speed,etc.,Otsu image segmentation algorithm based on fractional hybrid bat algorithm was proposed.Fractional order calculus,beetle antennae search and traditional bat algorithm were combined,and the genetic characteristics of fractional order calculus were used to balance the global search process of bats,thereby improving the optimization ability.Beetle antennae search was used to enrich the diversity of bat population during local search,the convergence of the algorithm was speeded up under high precision.The maximum inter-class variance of the two-dimensional Otsu algorithm was used as the fitness function for image threshold segmentation.Experimental results show that the improved algorithm improves the image segmentation accuracy and effect,and also has higher convergence speed and better robustness.
作者
梁远哲
马瑜
江妍
王原
马鼎
李霞
LIANG Yuan-zhe;MA Yu;JIANG Yan;WANG Yuan;MA Ding;LI Xia(School of Physics and Electronic-Electrical Engineering,Ningxia University,Yinchuan 750021,China)
出处
《计算机工程与设计》
北大核心
2021年第11期3091-3098,共8页
Computer Engineering and Design
基金
宁夏自然科学基金项目(NZ16009)
宁夏高等学校科学研究基金项目(NGY2016015)
2018年宁夏研究生教育教学改革研究与实践基金项目(YJG201811)。
关键词
最大类间方差法
分数阶微积分
蝙蝠算法
天牛须搜索
图像分割
maximum class variance method
fractional order
bat algorithm
beetle antennae search
image segmentation