dival.util.odl_noise_random_state module

Functions to create noise samples of different distributions.

This module is a copy of odl.phantom.noise that was modified to support random states instead of seeds.

dival.util.odl_noise_random_state.white_noise(space, mean=0, stddev=1, seed=None, random_state=None)[source]

Standard gaussian noise in space, pointwise N(mean, stddev**2).

Parameters

  • space (TensorSpace or ProductSpace) – The space in which the noise is created.

  • mean (space.field element or space element-like, optional) – The mean of the white noise. If a scalar, it is interpreted as mean * space.one(). If space is complex, the real and imaginary parts are interpreted as the mean of their respective part of the noise.

  • stddev (float or space element-like, optional) – The standard deviation of the white noise. If a scalar, it is interpreted as stddev * space.one().

  • seed (int, optional) – Random seed to use for generating the noise. For None, use the current seed. Only takes effect if random_state is None.

  • random_state (numpy.random.RandomState, optional) – Random state to use for generating the noise. For None, use the global numpy random state (i.e. functions in np.random).

Returns

white_noise

Return type

space element

See also

poisson_noise(), salt_pepper_noise(), numpy.random.normal()

dival.util.odl_noise_random_state.poisson_noise(intensity, seed=None, random_state=None)[source]

Poisson distributed noise with given intensity.

Parameters

intensity (TensorSpace or ProductSpace element) – The intensity (usually called lambda) parameter of the noise.

Returns

  • poisson_noise (intensity.space element) – Poisson distributed random variable.

  • seed (int, optional) – Random seed to use for generating the noise. For None, use the current seed. Only takes effect if random_state is None.

  • random_state (numpy.random.RandomState, optional) – Random state to use for generating the noise. For None, use the global numpy random state (i.e. functions in np.random).

Notes

For a Poisson distributed random variable \(X\) with intensity \(\lambda\), the probability of it taking the value \(k \in \mathbb{N}_0\) is given by

\[\frac{\lambda^k e^{-\lambda}}{k!}\]

Note that the function only takes integer values.

See also

white_noise(), salt_pepper_noise(), uniform_noise(), numpy.random.poisson()

dival.util.odl_noise_random_state.salt_pepper_noise(vector, fraction=0.05, salt_vs_pepper=0.5, low_val=None, high_val=None, seed=None, random_state=None)[source]

Add salt and pepper noise to vector.

Salt and pepper noise replaces random elements in vector with low_val or high_val.

Parameters

  • vector (element of TensorSpace or ProductSpace) – The vector that noise should be added to.

  • fraction (float, optional) – The propotion of the elements in vector that should be converted to noise.

  • salt_vs_pepper (float, optional) – Relative abundance of salt (high) vs pepper (low) noise. A high value means more salt than pepper noise.

  • low_val (float, optional) – The “pepper” color in the noise. Default: minimum value of vector. For product spaces the minimum value per subspace is taken.

  • high_val (float, optional) – The “salt” value in the noise. Default: maximuim value of vector. For product spaces the maximum value per subspace is taken.

  • seed (int, optional) – Random seed to use for generating the noise. For None, use the current seed. Only takes effect if random_state is None.

  • random_state (numpy.random.RandomState, optional) – Random state to use for generating the noise. For None, use the global numpy random state (i.e. functions in np.random).

Returns

salt_pepper_noise – vector with salt and pepper noise.

Return type

vector.space element

See also

white_noise(), poisson_noise(), uniform_noise()

dival.util.odl_noise_random_state.uniform_noise(space, low=0, high=1, seed=None, random_state=None)[source]

Uniformly distributed noise in space, pointwise U(low, high).

Parameters

  • space (TensorSpace or ProductSpace) – The space in which the noise is created.

  • low (space.field element or space element-like, optional) – The lower bound of the uniform noise. If a scalar, it is interpreted as low * space.one(). If space is complex, the real and imaginary parts are interpreted as their respective part of the noise.

  • high (space.field element or space element-like, optional) – The upper bound of the uniform noise. If a scalar, it is interpreted as high * space.one(). If space is complex, the real and imaginary parts are interpreted as their respective part of the noise.

  • seed (int, optional) – Random seed to use for generating the noise. For None, use the current seed. Only takes effect if random_state is None.

  • random_state (numpy.random.RandomState, optional) – Random state to use for generating the noise. For None, use the global numpy random state (i.e. functions in np.random).

Returns

white_noise

Return type

space element

See also

poisson_noise(), salt_pepper_noise(), white_noise(), numpy.random.normal()