AddNoise

MLModule

genre	Generators
author	MeVis Medical Solutions AG
package	MeVisLab/Standard
dll	MLBasicModules
definition	mlBasicModules.def
keywords	uniform, gaussian, white, random, disturbance, inhomogeneity, testing, segmentation, gray, gaussian, speckle, salt, pepper, checkerboard

Purpose

The AddNoise module produces noise from various distributions.

Usage

Connect an image, choose a distribution, and set the parameters. Note that only scalar input images are currently supported, i.e., adding noise to vector or tensor images is not supported.

Details

Random numbers are generated using the boost library and the lagged_fibonacci607 random number generator. The following noise distributions are available:

Uniform	Adds uniformly distributed noise to the image: I = I + noise
Gaussian	Adds Gaussian noise of mean m and variance v to the image: I = I + noise
Gaussian DTI	Adds complex Gaussian noise of variance v to the MR signal
Salt & Pepper	Adds salt and pepper noise to the image (intensity=amplitude or amplitude2 for density*100% of the voxels): I = I + noise
Speckle	Adds multiplicative noise (Gaussian white noise of variance v) to the image: I = I + I*noise

Tips

If just a noise image is desired, produce a constant image with the ConstantImage module and attach it to the AddNoise module.

Windows

Default Panel

Input Fields

A scalar image volume. Vector and tensor images are not yet supported.

input0

name: input0, type: Image

Output Fields

A noisy image.

output0

name: output0, type: Image

Parameter Fields

Field Index

Add Input: Bool	sigma (input image): Float
Density: Float	sigma (output image): Float
Gray Value 1 (Salt): Float
Gray Value 2 (Pepper): Float
Mean Value: Float
Noise Type: Enum
RandSeed (negative: unused): Integer
sigma (additive): Float

Visible Fields

Gray Value 1 (Salt)

name: amplitude, type: Float, default: 100

Add Input

name: addInput, type: Bool, default: TRUE

If checked, the voxel values of the input image are being added to the output (noise) voxel values.

RandSeed (negative: unused)

name: randSeed, type: Integer, default: -1

Sets the seed for the random number generator.

If a negative number is selected, a seed is chosen based on the current time.

Noise Type

name: noiseType, type: Enum, default: Uniform

Defines the type of noise that is being generated.

Values:

Title	Name	Description
Gaussian	Gaussian	Mean Value mean of Gaussian. Sigma sigma of Gaussian (sqrt of variance).
Gaussian Dti	GaussianDti	sigma (additive) sigma (σ) of Gaussian (sqrt of variance) used for adding complex Gaussian noise to the input image. When changing sigma (input image) or sigma (output image), the sigma (additive) is updated automatically. σadditive = √ (σoutput2- σinput2) sigma (input image) noise of the input image sigma (output image) (desired) noise of the output image image. If it is smaller than sigma of the input image, the additive sigma is set to 0.
Salt Pepper	SaltPepper	Gray Value 1 (Salt) salt value Gray Value 2 (Pepper) pepper value Density noise density; approximately density*100% of the voxels are affected
Speckle	Speckle	sigma sigma of Gaussian (sqrt of variance)
Checkerboard	Checkerboard

Mean Value

name: mean, type: Float, default: 0

sigma (additive)

name: sigma, type: Float, default: 1

sigma (input image)

name: sigmaIn, type: Float, default: 0

sigma (output image)

name: sigmaOut, type: Float, default: 0

Gray Value 2 (Pepper)

name: amplitude2, type: Float, default: 100

Density

name: density, type: Float, default: 0.050000001