摘要
提出了一种基于改进BP神经网络和粒子群优化算法(PSO)的图像滤波方法。该方法利用双曲正切形式的误差函数代替BP神经网络传统的最小均方误差函数(LMS),并将改进后的BP神经网络利用PSO算法优化,用来减小图像噪声对神经网络精度的影响以及避免神经网络陷入局部极小值点,从而提高神经网络去噪能力。实验结果表明,与传统滤波方法相比,该方法不仅能有效地滤除图像中的高斯噪声而且能很好地保护图像细节。
A new method for image noise reduction based on particle swarm optimization (PSO) and modified BP neural network is proposed.This method introduces BP neural network by utilizing hyperbolic tangent error function instead of mean square error function as its cost function,and then the modified BP neural network is optimized with PSO.The proposed method can minish the influence on the accuracy of BP neural network model which is controlled by image noise and avoid local infinitesimal obviously.Experimental re...
出处
《光电子.激光》
EI
CAS
CSCD
北大核心
2009年第3期406-409,共4页
Journal of Optoelectronics·Laser
基金
教育部新世纪优秀人才支持计划资助项目(NCET-05-0897)
关键词
图像滤波
BP神经网络
双曲正切函数
粒子群优化(PSO)
image filtering
BP neural network
hyperbolic tangent error function
particle swarm optimization(PSO)