基于阵列级联FHN神经元的弱信号随机共振复原研究  

Research on Weak Signal Recovery Based on Stochastic Resonance of Array-Cascade FHN Neurons

作　　者：陈金龙[1] 范影乐[1] 武薇[1] 高云园[1] 

机构地区：[1]杭州电子科技大学模式识别与图像处理实验室

出　　处：《航天医学与医学工程》2013年第4期317-322,共6页Space Medicine & Medical Engineering

基　　金：国家自然科学基金资助项目(60872090;61201300);浙江省自然科学基金资助项目(LY12F03006)

摘　　要：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 To propose a array-cascade FHN neurons model for weak signal recovery basing on the multi-input connection of neuron synapses and the continuous transfer characteristics of action potential. Methods To describe the correlation of image pixels in the neighborhood fully, the combination of raster scan and Hilbert scan was adopted to the two-dimensional image restoration. The indicator of peak signal-to-noise ratio was conducted to analyze the effect of low signal-to-noise ratio image restoration. Results The array-cas- cade FHN neurons model could reduce noise effectively, reinforce signal contour edges and details, make the signal sense of hierarchy stronger and improve the robustness to internal noise. Conclusion The stochastic res- onance mechanism based on array-cascade FHN neurons provides a new approach for weak signal recovery.

关 键 词：阵列级联FHN模型 随机共振 弱信号复原 

分 类 号：R857.3[医药卫生—航空、航天与航海医学] TP391[医药卫生—临床医学]