Learning Camera-Aware Noise Models

TL;DR

This paper introduces a data-driven, camera-aware noise model that learns from real-world sensor noise, outperforming traditional statistical models in accuracy and versatility for image processing tasks.

Contribution

The paper presents a novel generative noise model that captures camera-specific noise characteristics, enabling more realistic noise synthesis across different sensors.

Findings

Outperforms existing statistical noise models

Generates realistic, camera-specific noise

Works effectively across multiple camera sensors

Abstract

Modeling imaging sensor noise is a fundamental problem for image processing and computer vision applications. While most previous works adopt statistical noise models, real-world noise is far more complicated and beyond what these models can describe. To tackle this issue, we propose a data-driven approach, where a generative noise model is learned from real-world noise. The proposed noise model is camera-aware, that is, different noise characteristics of different camera sensors can be learned simultaneously, and a single learned noise model can generate different noise for different camera sensors. Experimental results show that our method quantitatively and qualitatively outperforms existing statistical noise models and learning-based methods.