Stochastic Resonance Optimization Model And Its Application Research In Image Enhancement

Under the condition of appropriate system parameters and noise intensity,the mechanism of stochastic resonance will drive the noise energy to transfer to the weak signal,which changes the viewpoint of weak signal detection.But now most of the research focuses on the verification of the phenomenon of stochastic resonance in different nonlinear systems.Besides,the study of application for the weak signal enhancement is usually taken strategy as a fixed or semi fixed model structure and parameters,and the performance of optimal stochastic resonance of a specific condition can be achieved by means of an artificial adjustment of some parameters.So,we use the swarm optimizational algorithm of quantum particle to optimize the parameters of the model simultaneously in this paper,to avoid the drawbacks by experience and manually adjusting the system' parameter separately.For improving the efficiency and performance of weak signal enhancement effectively,it has a good significance.On the other hand,a model of multilayer stochastic resonance is constructed basing on the suppression of synaptic neuron interconnections in this paper,making up the shortfalls of globally approximating optimizational problem of detection of multiple contrasts of weak signal by the traditional method,and extracting application of multiple layer edge in biological colony image.Finally,a new method of hierarchical stochastic resonance based on multi-scale wavelet decomposition is proposed in this paper,making the overall optimizational problem of mixed weak signal being converted to a hierarchical optimization problem at different scales,as a result that the enhancement performance of stochastic resonance is improved significantly.The main work and research results of this paper are as follows:(1)The traditional stochastic resonance in the image processing are mostly experiential manual adjusting parameters,which would lead to lower efficiency,so a method of combining stochastic resonance with swarm optimization algorithm of quantum particle is proposing in this paper.Firstly,a two stage parallel of stochastic resonance system is constructed which improves the stability and the low pass filter performance of the stochastic resonance system.Secondly,I have compared the effect of the swarm optimization algorithm of quantum particle and the common adaptive swarm optimization algorithm of particle combined with the stochastic resonance.Finally,we extend the stochastic resonance adaptive parameter optimization algorithm from one-dimension to two-dimensional medical signal to enhance the contrast research.The experimental results show that the proposed method can effectively enhance the contrast of low dose CT images.(2)In multiple contrast regions of the image,single layer stochastic resonance is difficult to obtain the information of image edge feature completely,and an edge detection method for weak signal detection in multi-layered suppression of synaptic random resonance is proposed.Firstly,a series-parallel connection FHN neuron structure model is established,and the quantum particle swarm optimization algorithm is adopted as the parameter optimization tool.Secondly,detecting the intensity of the one dimensional mutation signal,and then using the mean time error as the evaluation index to detect the actual mutation edge points and the extracted pulse points.Finally,the edge detection is applied to multiple contrast regions of the two-dimensional biological colonies images and maximum Shannon entropy is taken as a quantitative evaluation index.The experimental results show that the edge details of this method are rich,and the level is more stronger.(3)The method of traditional stochastic resonance transfer noise energy in the spatial domain is mainly as a global meaning stochastic resonance process,ignoring the noise and image main signal in different frequency bands.So,according to the characteristics of the image signal and the noise signal in the frequency domain,the integration of stochastic resonance and multiple scale frequency domain decomposition is constructed.Firstly,to obtain the sub signal of different frequency levels,the image signal is decomposed by wavelet multi-scale,so that the main gray information of the image concentrate in the low frequency section,and the details information of the image and noise mainly distribute in the middle and high frequency sections.Secondly,using quantum particle swarm optimization to optimize the parameters of the bistable stochastic resonance system which enhances the sub information of the decomposed image.Finally,the each sub signal results are transformed by wavelet inverse to obtain the final image.Experimental results show that the proposed method in this paper has a good effect of image denoising on the low quality biomedical images,and improves the overall quality of the image effectively.