2 Hits in 4.0 sec

Extending Stein's unbiased risk estimator to train deep denoisers with correlated pairs of noisy images [article]

Magauiya Zhussip, Shakarim Soltanayev, Se Young Chun
2019 arXiv   pre-print
Recently, Stein's unbiased risk estimator (SURE) has been applied to unsupervised training of deep neural network Gaussian denoisers that outperformed classical non-deep learning based denoisers and yielded  ...  per image for Noise2Noise).  ...  Stein's unbiased risk estimator (SURE) based training method for Gaussian denoisers was proposed to train DNNs with a set of a single noise realization per image, but it was limited to Gaussian denoising  ... 
arXiv:1902.02452v2 fatcat:7n354pswhredxkigyvpubbu54y

Noise2Self: Blind Denoising by Self-Supervision [article]

Joshua Batson, Loic Royer
2019 arXiv   pre-print
This framework generalizes recent work on training neural nets from noisy images and on cross-validation for matrix factorization.  ...  We propose a general framework for denoising high-dimensional measurements which requires no prior on the signal, no estimate of the noise, and no clean training data.  ...  Thank you to Jack Kamm for discussions on Gaussian Processes and shrinkage estimators.  ... 
arXiv:1901.11365v2 fatcat:n672sj7fung27dexfzxlx43zga