Maximum-likelihood estimation in ptychography in the presence of Poisson-Gaussian noise statistics

We introduce a novel method for maximum-likelihood estimation in ptychography to address the challenge posed by mixed Poisson-Gaussian noise statistics. By integrating a loss function that accounts for both noise sources in computational image retrieval, our approach significantly improves image reconstruction quality under low signal-to-noise ratio conditions. Experimental and numerical data confirm the advantage of our method over traditional approaches that consider only Poissonian noise. This advancement promises enhanced performance in computational imaging applications, particularly in situations where accurate noise modeling is crucial.