结合PCNN和局部维纳滤波的图像去噪  

An Image De-Noising Algorithm Based on PCNN and Local Wiener Filter

作　　者：宫霄霖[1] 毛瑞全[1] 

机构地区：[1]天津大学电子信息工程学院,天津300072

出　　处：《北京邮电大学学报》2011年第5期67-70,共4页Journal of Beijing University of Posts and Telecommunications

摘　　要：提出了一种基于脉冲耦合神经网络(PCNN)的图像去噪算法,利用PCNN的同步脉冲特性对图像小波系数进行局部加窗修正,从而对信号方差进行更好地估计,以利用局部维纳滤波进行去噪.同时,根据各个像素间耦合特性的不同,提出了自适应连接系数,更好地反映了像素间的耦合关系,利于信号方差的估计.实验结果表明,该算法较维纳滤波和NeighShrink算法均有较高的峰值信噪比,视觉效果更好.A A new de-noising algorithm based on pulse coupled neural network （PCNN） system and local Wiener filter is presented. The mechanism of synchronous pulse of PCNN is employed to revise the image wavelet coefficients, according to the feather of different image pixels. Also, an adaptive linking coefficient/3 is proposed. Based on the improved wavelet coefficients, the local Wiener filter is used to estimate signal variance better. Experiments show the proposed scheme can get higher peak signal-to-noise ratio than NeighShrink and Wiener algorithm, and better vision effect.

关 键 词：脉冲耦合神经网络 自适应连接系数 图像去噪 局部加窗 噪声检测 

分 类 号：TN391[电子电信—物理电子学]