UniformNoise

class deepinv.physics.UniformNoise(a=0.1, rng: Generator | None = None)[source]

Bases: NoiseModel

Uniform noise \(y = x + \epsilon\) where \(\epsilon\sim\mathcal{U}(-a,a)\).


Examples:

Adding uniform noise to a physics operator by setting the noise_model attribute of the physics operator:

>>> from deepinv.physics import Denoising, UniformNoise
>>> import torch
>>> physics = Denoising()
>>> physics.noise_model = UniformNoise()
>>> x = torch.rand(1, 1, 2, 2)
>>> y = physics(x)
Parameters:
  • a (float) – amplitude of the noise.

  • rng (torch.Generator (Optional)) – a pseudorandom random number generator for the parameter generation.

forward(x, a=None, seed: int | None = None, **kwargs)[source]

Adds the noise to measurements x

Parameters:
  • x (torch.Tensor) – measurements

  • a (float, torch.Tensor) – amplitude of the noise. If not None, it will overwrite the current noise level.

  • seed (int) – the seed for the random number generator, if rng is provided.

Returns:

noisy measurements

update_parameters(a=None, **kwargs)[source]

Updates the amplitude of the noise.

Parameters:

a (float, torch.Tensor) – amplitude of the noise.

Examples using UniformNoise:

Patch priors for limited-angle computed tomography

Patch priors for limited-angle computed tomography