基于FitzHugh-Naguno神经元随机共振机制的图像复原

应用随机共振机制,借助噪声能量实现图像复原,改善低信噪比图像的输出质量.通过分析FitzHugh-Nagumo(FHN)神经元的阈上随机共振机理,以及在相平面上的阈值工作特性,对FHN神经元模型进行约简,以峰值信噪比(PSNR)为图像复原的评价函数,提出基于随机共振的自适应最优图像复原算法.以含噪mountain彩色图像和LED芯片图像为实验对象,与均值滤波、维纳滤波等图像复原算法进行仿真对比研究.结果表明:随机共振方法较好地抑制了噪声、恢复了图像细节和色彩信息,且随着噪声的增强,随机共振方法复原图像的峰值信噪比变化较小,该方法具有较好的鲁棒性.

Image restoration based on stochastic resonance mechanism of FitzHugh-Nagumo neuron

Based on stochastic resonance(SR) mechanism, the quality of low signal noise ratio image was improved by adding noise energy. According to analyzing suprathreshold stochastic resonance and threshold working characteristic of Fitzhugh Nagumo(FHN) neuron model at phase space, the model was simplified. By choosing the peak signal to noise ratio(PSNR) as estimation function of the image restoration, a self adaptive optimized algorithm of image restoration processing based on RS mechanism was introduced. A noisy mountain color image, LED chip image and so on, were chosen as experimental objects, to compare simulation result of image restoration with different algorithm, stochastic resonance restoration, mean filter restoration and Wiener filter restoration. The results indicate that SR image restoration method is better in noise inversion, image detail restoration and color information restoration. Along with the increasing of noise intensity, the peak signal to noise ratio of SR restoration images change less than others. This method has good robustness.