PSO优化PCNN模型在图像分割中的性能分析  

Performance Analysis of Image Segmentation Based on PSO Optimized PCNN Models

作　　者：杨旭 何鸿宇[3] 李金锁 廖源 周同驰[3] YANG Xu;HE Hongyu;LI Jinsuo;LIAO Yuan;ZHOU Tongchi(Things Application Engineering Technology Research Center of Henan Industrial Internet,Nanyang 473005,China;Henan Institute of University and Technology,Nanyang 473005,China;Zhongyuan University of Technology,Zhengzhou 451191,China)

机构地区：[1]河南省工业物联网应用工程技术研究中心,河南南阳473005 [2]河南工业职业技术学院,河南南阳473005 [3]中原工学院,河南郑州451191

出　　处：《机械与电子》2023年第11期43-48,共6页Machinery & Electronics

基　　金：教育部第三批"云数融合科教创新"基金项目(2018A10004);河南省高校学校重点科研项目(20B120004);中国纺织工业联合会指导项目(2018107)。

摘　　要：脉冲耦合神经网络模型参数众多,在应用时通常根据经验或实验设置。针对模型参数影响应用等问题,采用群体智能优化算法PSO优化简单脉冲耦合神经网络模型的关键参数,如耦合系数、时间衰减因子和脉冲输出的乘积系数。通过仿真实验,评价智能优化算法学习模型参数的有效性,比较分析了不同参数PCNN模型在图像分割中获得的轮廓、细节和纹理等存在差异性。The numerous parameters of the pulse coupled neural network model affects the application,and the researchers usually set them according to experience or experiment.To relieve this problem,the key parameters of the simple pulse coupled neural network model,such as coupling coefficient,time attenuation factor,product coefficient of pulse output,are optimized by swarm intelligent optimization algorithm PSO.Through experiments,the advantage of PSO optimized PCNN model in the field of image segmentation is verified,and the differences of contours,details and textures obtained by different parameter PCNN models are compared and analyzed.

关 键 词：粒子群优化 脉冲耦合神经网络 图像分割 性能分析 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]