Practical Poissonian-Gaussian Noise Modeling and Fitting for Single-Image Raw-Data

By Alessandro Foi, Mejdi Trimeche, Vladimir Katkovnik, Karen O. Egiazarian. IEEE Transactions on Image Processing, 2008

990 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Electrical and Electronic Engineering, Media Technology

View Paper: https://www.semanticscholar.org/paper/8b709c9feabcec4f2b6eea92266c198c9163d4ec

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Electrical and Electronic Engineering

In Media Technology

Citations in the corpus

Patch-Based Nonlocal Functional for Denoising Fluorescence Microscopy Image Sequences (2010, background): “This transform was further generalized to Poisson-Gaussian noise [5], with satisfying results if the number of counts is large enough and more recently for “clipped” (under- and over-exposure) raw-data [6].”
The Grid Method for In‐plane Displacement and Strain Measurement: A Review and Analysis (2016, background): “This is however a rough assumption since actual camera sensor noise is Poisson-Gaussian and heteroscedastic (or signaldependent): the higher the brightness, the higher the noise level [185], and noise variance in images may reasonably be modeled with an affine function of the grey level [186, 187, 1”
Unprocessing Images for Learned Raw Denoising (2018, background): “The overall noise can more accurately be modeled as containing both Gaussian and Poissonian signal-dependent components [14] or as being sampled from a heteroscedastic Gaussian where variance is a function of intensity [20].”
CycleISP: Real Image Restoration via Improved Data Synthesis (2020, background): “Real camera noise is fundamentally different from AWGN, thereby causing a major challenge for deep CNNs [6, 22, 24].”
A High-Quality Denoising Dataset for Smartphone Cameras (2018, methodology): “For our noisy images, we report the NLF parameters provided by the camera device through the Android Camera2 API [15], which we found to be accurate when matched with [14].”
Deep joint demosaicking and denoising (2016, methodology): “Processing linear data Khashabi et al. [2014] suggest that demosaicking should be evaluated on raw RGB data with an affine noise model [Foi et al. 2008; Hasinoff et al. 2010].”
Benchmarking Denoising Algorithms with Real Photographs (2017, methodology): “Gaussian [13, 21], yet even seemingly minor details of the synthetic noise process, such as whether the noisy values are rounded to integers, can have a significant effect on the relative performance of methods [6, 33].”
Content-Adaptive Steganography by Minimizing Statistical Detectability (2016, methodology): “According to [15], [17], these values are fairly typical for a variety of imaging sensors at ISO 200.”
A Holistic Approach to Cross-Channel Image Noise Modeling and Its Application to Image Denoising (2016, positive): “In [6], Foi et al. proposed a Poissonian-Gaussian noise model for single-image raw data to consider signal-dependent and signal-independent noise components separately.”