The Research Of Image’s Two-dimensional Stochastic Resonance

The research studies the nonlinear bistable system. Based on analyzing bistable system structure, image noise and image denoising, the paper takes a close study into the application of nonlinear stochastic resonance in image processing, and further broadens its application range.On the basis of the one-dimensional signal stochastic resonance, the paper has presented the method of image stochastic resonance:two-dimensional stochastic resonance. The two-dimensional stochastic resonance can be accomplished by twice one-dimensional stochastic resonances. First process the row pixel with stochastic resonance, and then process the column pixel of the first processing result, in the end normalize and homogenize the second processing result. Generally the study object of stochastic resonance must fulfill the requirement of small parameters:signal amplitude, frequency and noise intensity must be far less than 1, therefore the paper proposes the method of amplitude normalization and calculation step length rescaling to ensure the image signal’s amplitude and frequency much smaller than 1. Simulation shows that two-dimensional stochastic resonance is able to remove the noise from the image effectively and enhance the quality of the image. Compared with traditional methods, the two-dimensional stochastic resonance is proved to be better.In order to evaluate the image processing result, the paper puts forward two objective assessments:the similarity coefficient and the deviation coefficient, which reflect the stochastic resonance’s effectiveness in image processing by different perspective. Based on the study of the two coeffeicient’s variation trend along with bistable system parameters and re-scaling calculation step length, the research gained the trend of image processing quality with these variables, which laid the basis for the adaptive parameter optimization of the SR image processing...