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

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.