PatchCovarianceNoiseEstimator

class deepinv.models.PatchCovarianceNoiseEstimator

Bases: Module

Pach Covariance Gaussian noise level estimator.

This method was initially proposed in Chen et al.[1]. Given a noisy image \(y = x + n\) where \(n \sim \mathcal{N}(0, \sigma^2)\), this estimator computes an estimate of \(\sigma\) based on the eigenvalues of the covariance matrix of image patches.

Warning

This estimator assumes that the noise in the corrupted image follows a Gaussian distribution. It may not perform well if the noise distribution deviates significantly from Gaussian, or if the image lacks sufficient homogeneous regions for reliable patch statistics.

---

Examples:

>>> import deepinv as dinv
>>> from deepinv.models import PatchCovarianceNoiseEstimator
>>> rng = torch.Generator('cpu').manual_seed(0)
>>> noise = dinv.physics.GaussianNoise(0.1, rng=rng)(torch.zeros(1, 1, 256, 256))
>>> noise_estimator = PatchCovarianceNoiseEstimator()
>>> sigma_est = noise_estimator(noise)
>>> print(sigma_est)
tensor([0.0995])

---

References:

[1]

Guangyong Chen, Fengyuan Zhu, and Pheng Ann Heng. An efficient statistical method for image noise level estimation. In Proceedings of the IEEE international conference on computer vision, 477–485. 2015.

static estimate_noise(x, pch_size=8)

Estimates noise level in image im.

Parameters:

• x (torch.Tensor) – input image

• pch_size ((int, int)) – patch size

Returns:

(torch.Tensor) estimated noise level

Return type:

Tensor

forward(x)

Forward pass.

Parameters:

x (torch.Tensor) – input image

Returns:

(torch.Tensor) estimated noise level

Return type:

Tensor

Examples using PatchCovarianceNoiseEstimator:

Blind denoising with noise level estimation

Blind denoising with noise level estimation