摘要
为了降低脉冲耦合神经网络(pulse coupled neural network,PCNN)参数设置的复杂度,提出了一种利用哈里斯鹰优化算法(Harris Hawk optimization algorithm,HHO)搜索PCNN参数的图像自动分割方法。一方面,在不影响分割效果的情况下,减少了PCNN的参数个数;另一方面,HHO算法具有收敛速度快、全局搜索能力强的特点,能够快速、准确地搜索到PCNN相应参数。引入图像熵作为适应度函数,选取脑部MRI图像进行实验,通过精度、召回率和dice,比较了HHO结合PCNN与几种不同搜索机制的优化算法结合PCNN的分割性能,仿真实验结果表明,提出的方法有较高的分割精度和较强的鲁棒性,具有较高的工程实用价值。
In order to reduce the complexity of setting parameters of pulse coupled neural network(PCNN),an au?tomatic image segmentation method is proposed,which uses Harris Hawk optimization algorithm(HHO)to search PCNN parameters.On the one hand,without affecting the segmentation effect,the number of PCNN parameters is reduced.On the other hand,Harris Hawk optimization algorithm has the characteristics of fast convergence and strong global search ability,and can quickly and accurately search the corresponding parameters of PCNN.The im?age entropy is introduced as the fitness function,and brain MRI images are selected for experiment.The segmenta?tion performance of HHO combined with PCNN is compared with several optimization algorithms that combine PC?NN and have different search mechanisms in aspects of precision,recall and dice.The simulation results show that the proposed method has higher segmentation accuracy and robustness,and has higher engineering application value.
作者
贾鹤鸣
康立飞
孙康健
彭晓旭
李瑶
姜子超
JIA Heming;KANG Lifei;SUN Kangjian;PENG Xiaoxu;LI Yao;JIANG Zichao(College of Mechanical and Electrical Engineering,Northeast Forestry University,Harbin 150040,China)
出处
《应用科技》
CAS
2019年第4期16-20,25,共6页
Applied Science and Technology
基金
国家自然科学基金项目(51609048)
东北林业大学横向课题项目(43217002,43217005,43219002)
关键词
图像分割
哈里斯鹰优化算法
脉冲耦合神经网络
熵
评价标准
参数自适应
智能优化算法
神经网络
image segmentation
Harris hawk optimization algorithm
pulse coupled neural network
entropy
evaluation criteria
parameter adaptation
intelligent optimization algorithm
neural network