dsp.Normalizer System object

Package: dsp

Vector normalization along specified dimension


The Normalizer object performs vector normalization along rows, columns, or specified dimension.

To perform vector normalization:

  1. Define and set up your normalization object. See Construction.

  2. Call step to perform vector normalization according to the properties of dsp.Normalizer. The behavior of step is specific to each object in the toolbox.


H = dsp.Normalizer returns a normalization object, H, that normalizes the input over each column by the squared 2-norm of the column plus a bias term of 1e-10 used to protect against divide-by-zero.

H = dsp.Normalizer('PropertyName',PropertyValue, ...) returns a normalization object, H, with each property set to the specified value.



Type of normalization to perform

Specify the type of normalization to perform as 2-norm or Squared 2-norm. The 2-norm mode supports floating-point signals only. The Squared 2-norm supports both fixed-point and floating-point signals. The default is Squared 2-norm.


Real number added in denominator to avoid division by zero

Specify the real number to add in the denominator to avoid division by zero. The default is 1e-10. This property is tunable.


Dimension to operate along

Specify whether to normalize along Column , Row, or Custom. The default is Column.


Numerical dimension to operate along

Specify the one-based value of the dimension over which to normalize. The value of this parameter cannot exceed the number of dimensions in the input signal. This property applies when Dimension property is Custom. The default is 1.

 Fixed-Point Properties


cloneCreate normalization object with same property values
getNumInputsNumber of expected inputs to step method
getNumOutputsNumber of outputs of step method
isLockedLocked status for input attributes and nontunable properties
releaseAllow property value and input characteristics changes
stepNormalize input along specified dimension


Normalize a matrix:

 hnorm = dsp.Normalizer;
 x = magic(3);
 y = step(hnorm, x);


This object implements the algorithm, inputs, and outputs described on the Normalization block reference page. The object properties correspond to the block parameters, except:

  • Treat sample-based row input as column — The block allows you to input a row vector and normalize the row vector as a column vector. The normalization object always normalizes along the value of the Dimension property.

  • The normalization object does not support the Minimum and Maximum options for data output.

Introduced in R2012a

Was this topic helpful?