论文部分内容阅读
讨论了电子倍增CCD(EMCCD)图像的噪声来源及其统计特性,建立了混合泊松-高斯噪声分布模型。针对混合泊松-高斯噪声分布模型的极大似然函数难以求解的问题,对噪声模型进行了适当的初始化设置,利用期望最大化算法对噪声模型进行参数估计,有效实现了噪声参数的极大似然估计。Monte Carlo仿真结果及实验结果表明,期望最大化算法估计性能较好,对混合泊松-高斯分布有较好的拟合效果,能得到较高精度的参数估计值。
The sources of noise and its statistical properties of electron multiplying CCD (EMCCD) images are discussed. A mixed Poisson - Gaussian noise distribution model is established. Aiming at the difficulty of solving the maximum likelihood function of the mixed Poisson-Gaussian noise distribution model, the noise model is properly initialized, and the noise maximization algorithm is used to estimate the parameters of the noise model. Likelihood estimation. Monte Carlo simulation results and experimental results show that the expectation maximization algorithm has better estimation performance, good fitting effect on the mixed Poisson - Gaussian distribution, and higher accuracy of the parameter estimation.