Poisson2Gaussian: Noise Gaussianization to Enhance Image Denoising
2026-06-22 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors studied the noise in photos caused by the random nature of light, called Poisson noise, which is tricky for computer programs to clean because it changes depending on the signal and isn't evenly distributed. They created a new method called Poisson2Gaussian (P2G) that transforms this complicated noise into simpler Gaussian noise, which is easier for programs to handle. Their approach does not need clean example photos or exact noise details and works well with different noise-removal methods. Tests show that P2G generally improves the clarity of images, especially when the noise is very different from normal Gaussian noise.
Poisson noiseGaussian noisedenoisingheteroscedasticitysignal dependencyprobability density matchingphotonic noiseimage processingPSNRnoise modeling
Authors
Xirou Zhou, Zijing Xu, Yibo Qu, Qi Zhang, Xiaowan Hu, Xinyang Li
Abstract
The quantum nature of light determines the inherent Poisson stochasticity of photon detection, which is ubiquitous in photography, microscopy, and astronomy. However, our controlled numerical studies reveal that the signal-dependency, heteroscedasticity, and statistical asymmetry of Poisson-mixed noise make it challenging for existing denoisers to learn. In contrast, i.i.d. Gaussian noise, with its statistical independence and symmetric distribution, is easier to model for networks. To address this gap, we propose Poisson2Gaussian (P2G), a noise Gaussianization method that explicitly converts complex real-world noise to i.i.d. Gaussian noise via probability density matching beyond low-order moments. We also design an unbiased denoising framework that synergizes P2G with downstream denoisers, ensuring convergence to the underlying signal without requiring paired clean data or explicit noise parameters. Extensive experiments demonstrate that P2G consistently achieves state-of-the-art performance across diverse datasets. In challenging scenarios where noise strongly deviates from Gaussian statistics, our method improves the PSNR by up to 0.75 dB. Notably, P2G is architecture-agnostic and can provide universal improvements for various denoisers. The source code will be publicly available.