基于脉冲耦合神经网络的椒盐噪声滤波

Salt and pepper noise image filtering method using PCNN

基于脉冲耦合神经网络,提出了一种有效的椒盐噪声图像滤波算法。首先利用PCNN相似群神经元同步发放脉冲的特性检测噪声,并给出了神经元参数的估计方法。然后考虑到噪声点应和最近的非噪声点最相似,提出了一种扩展窗口中值滤波算法对噪声点进行滤波。仿真表明,本文提出的方法对不同强度的噪声图像均体现了优异的滤波性能,和相关的中值滤波算法相比也体现了相当明显的优势。

Based on Pulse Coupled Neural networks, an effective salt and pepper noise image filtering method is proposed. Synchronous pulses were burst by using the similar groups of neurons in a PCNN, whereby the noise pixels are detected;and the neuron parameter-estimation method was given. Then it is considered that a noise pixel has the most similar with neighbor non-noise pixels, a filtering method called extended window median filter was put forward, which filtered the noise in a noise image. Simulation results show that the proposed method has excellent filtering performance for the noise images of different noise intensity, and has the more obvious advantage than the corresponding median filters method.