Research Of Emccd Parameter Estimation And Noise Filtering Algorithm

The high sensitivity, high-quantum efficiency, and low noise make the EMCCD having a unique advantage in low-light imaging. However, in the conditions of low-light, the brightness and contrast of the image is relatively low. Enhancing the multiplication gain does help make the picture clearer, bringing out more picture noise at the same time. The best operating performance of the low-light imaging system is largely determined by the noise level, therefore, according to image noise sources and their statistic characteristics of the EMCCD, establishing a suitable noise distribution model for image processing and studying the estimation method for noise parameters will have an important theoretical significance and have broad application prospects for designing effective filters to suppress noise and reducing the influence of noise for image restoration.In this paper, we analyze the various noise sources and their statistical characteristics of the EMCCD image, the mixed Poisson-Gaussian distribution is used to establish the EMCCD noise distribution model. The moment estimation method, expectation maximization method and the Gauss-Newton method is used to complete the parameter estimation of the EMCCD noise distribution model. Then we discussed the specific estimation steps and implementation methods of the three algorithms based on mixed Poisson-Gaussian distribution models. Monte Carlo simulation results demonstrate the effectiveness of the proposed algorithm. Acquiring series of images from the EMCCD when there is no signal input, experimental results demonstrate that parameter estimation values agree fairly well with the EMCCD camera parameters, which will be benefit for the research of EMCCD noise suppression and image restoration.The EMCCD image noise parameter values can be obtained by using the three parameter estimation algorithms studied in this paper, the corresponding mixed Poisson-Gaussian noise can then be simulated, the wavelet semi-soft threshold algorithm is used to filter the EMCCD picture noise. Simulation results show that, wavelet semi-soft threshold filtering algorithm can effectively filter out the noise, restore a clear image, and can retain the image detail and edge information.