图像修复RBF模型的Kalman改进算法  

Improved Image Restoration RBF Model Based on the Kalman Filter

作　　者：游培寒[1] 毕笃彦[1] 毛柏鑫[1] 

机构地区：[1]空军工程大学工程学院,陕西西安710038

出　　处：《小型微型计算机系统》2005年第4期676-679,共4页Journal of Chinese Computer Systems

摘　　要：利用Kalman滤波器训练图像修复RBF网络模型的权值与基函数的中心,由于利用滤波器测量噪声的协方差阵实现整个模型的图像降噪功能,新算法边缘修复能力较强.本文通过仿真,检验了新模型的降噪能力;结果证明,新算法较一般的维纳滤波器,均值滤波降噪能力明显提高.This paper readopts a new kind of Kalman filter to train the Radial Basis Function (RBF) neural networks weight and the centers of the base functions, through setting the measurement noise s covariance matrix, the new algorithm realize noise-filtering capacity of the model, the remarkable character of the new algorithm is its powerful boundary restoration capacity;Through the simulation, this paper compare the new algorithm and the wiener and mean filter algorithm, the result shows that the new algorithm improve the models performance perfectly in the increment of the signal noise ratio aspects.

关 键 词：RBF 神经网络 KALMAN滤波器 ISNR 

分 类 号：TP317[自动化与计算机技术—计算机软件与理论]