Research On Signal Enhancement Based On FHN Neurons And Adaptive Stochastic Resonanc

Noise is usually considered to hinder the acquisition of original information,so the research of signal enhancement is based on the theory of noise suppression.Stochastic resonance(SR)theory subverted people’s cognition of noise.Noise does not only play a role of interference in signal acquisition,but an appropriate amount of noise can be converted into useful information to enhance the original input signal.Stochastic resonance has been used in many fields,especially in neurons.In this paper,the nonlinear system of Fitzhugh-Nagumo(FHN)neurons is combined with the stochastic resonance effect,and an adaptive algorithm is designed.Aiming at the problems of weak signal enhancement and low-brightness image contrast enhancement,the FHN neuron model based on adaptive stochastic resonance is proposed and the main work of this paper are as follows:(1)In this paper,a Fitzhugh-Nagumo neuronal pooling network model based on adaptive stochastic resonance is proposed.The different sizes of FHN neuronal nodes are used to enhance weakly periodic and aperiodic signals.The number of FHN neuron nodes and noise intensity were changed to analyze the influence of FHN neuron node number and noise intensity on weak periodic signals and aperiodic signals.Experimental results show that the proposed method has significant enhancement effect on both weak periodic signals and aperiodic signals,and the more FHN neuron nodes,the better the enhancement effect.(2)In this paper,a new method for solving optimal parameters of nonlinear systems is presented.The noise disturbance of original low-brightness image is eliminated by adaptive algorithm and the optimal parameters of FHN neural system are obtained.In this paper,The FHN neuron pooling network model based on adaptive stochastic resonance is extended to two-dimensional image processing,and the Kalman and least mean square(Kalman-LMS)adaptive algorithm designed in this paper is used to eliminate noise disturbance in input low-contrast images.The least square estimation(LSE)adaptive algorithm is used to obtain.The optimal parameters of the FHN nonlinear system.Experimental results show that the image processed by this model can effectively eliminate the noise disturbance.The model also provides a new method for solving the optimal parameter value of nonlinear system.(3)In this paper,an iterative model of FHN neuron based on adaptive stochastic resonance is proposed to enhance the color low-contrast images through adaptive iterative nonlinear equations.The size of the original low-contrast image is changed to analyze the influence of the image size on the enhancement effect.The optimal response is measured jointly by the relative contrast enhancement factor and the perceptual quality measurement.Experimental results show that the larger the image size,the more significant the enhancement effect.The FHN neuron iteration model is extended to self-synaptic FHN neuron and coupled FHN neuron nonlinear system,and the comparison between the three neuron iteration models proposed in this paper and existing methods under the same size is analyzed.However,compared with the existing methods,the contrast and the visual perception of the images are enhanced significantly by the model proposed in this paper.