论文部分内容阅读
目的基于神经元的突触多输入连接以及动作电位的连续传递特性,提出一种阵列级联FHN神经元模型,用于实现弱信号的复原。方法采用光栅扫描和Hilbert扫描相结合的方法对二维图像进行降维,以充分反映图像像素在邻域上的关联性,并基于峰值信噪比指标对低信噪比图像复原的效果进行分析。结果阵列级联FHN神经元模型能够有效抑制噪声,凸显信号轮廓边缘与细节,使信号层次感更强,同时对内噪声具有较强的鲁棒性。结论基于阵列级联FHN模型的随机共振机制将为弱信号复原提供一种新的思路。
OBJECTIVE Based on the synaptic multiple input connections of neurons and the continuous transfer of action potentials, an array-based FHN neuron model is proposed for the recovery of weak signals. Methods A two-dimensional image was reduced by using a combination of raster scan and Hilbert scan to fully reflect the correlation of image pixels in the neighborhood, and the effect of low signal-to-noise ratio image restoration based on peak signal to noise ratio . Results The array-cascaded FHN neuron model can effectively suppress the noise, highlight the edge and detail of the signal profile, enhance the signal level sense and have strong robustness to internal noise. Conclusion The stochastic resonance mechanism based on array cascaded FHN model will provide a new idea for weak signal recovery.