This website requires JavaScript.

PoGaIN: Poisson-Gaussian Image Noise Modeling from Paired Samples

Nicolas B\"ahlerMajed El Helou\'Etienne ObjoisKaan Okumu\c{s}Sabine S\"usstrunk
Oct 2022
摘要
Image noise can often be accurately fitted to a Poisson-Gaussiandistribution. However, estimating the distribution parameters from only a noisyimage is a challenging task. Here, we study the case when paired noisy andnoise-free samples are available. No method is currently available to exploitthe noise-free information, which holds the promise of achieving more accurateestimates. To fill this gap, we derive a novel, cumulant-based, approach forPoisson-Gaussian noise modeling from paired image samples. We show its improvedperformance over different baselines with special emphasis on MSE, effect ofoutliers, image dependence and bias, and additionally derive the log-likelihoodfunction for further insight and discuss real-world applicability.
展开全部
图表提取

暂无人提供速读十问回答

论文十问由沈向洋博士提出,鼓励大家带着这十个问题去阅读论文,用有用的信息构建认知模型。写出自己的十问回答,还有机会在当前页面展示哦。

Q1论文试图解决什么问题?
Q2这是否是一个新的问题?
Q3这篇文章要验证一个什么科学假设?
0
被引用
笔记
问答