cvr::noise Class Reference
[Non-linear filtering]

#include <cvrNoise.h>

Inheritance diagram for cvr::noise:

Collaboration diagram for cvr::noise:

List of all members.

Classes

class parameters

The parameters for the class noise. More...

Public Member Functions

noise ()

noise (const parameters &par)

noise (const noise &other)

virtual ~noise ()

bool apply (fmatrix &srcdest) const

bool apply (dmatrix &srcdest) const

bool apply (imatrix &srcdest) const

bool apply (matrix< ubyte > &srcdest) const

bool apply (fvector &srcdest) const

bool apply (dvector &srcdest) const

bool apply (ivector &srcdest) const

bool apply (const fmatrix &src, fmatrix &dest) const

bool apply (const dmatrix &src, dmatrix &dest) const

bool apply (const imatrix &src, imatrix &dest) const

bool apply (const matrix< ubyte > &src, matrix< ubyte > &dest) const

bool apply (const fvector &src, fvector &dest) const

bool apply (const dvector &src, dvector &dest) const

bool apply (const ivector &src, ivector &dest) const

noise & copy (const noise &other)

noise & operator= (const noise &other)

virtual const std::string & name () const

virtual noise * clone () const

virtual noise * newInstance () const

const parameters & getParameters () const

bool updateParameters ()

bool setNoiseGenerator (univariateContinuousDistribution &generator)

univariateContinuousDistribution * getNoiseGenerator ()

Protected Member Functions

parameters & getParameters ()

Protected Attributes

univariateContinuousDistribution * generator_

uniformDiscreteDistribution uniform_

Detailed Description

Class noise.

This class is used to add noise to an image. In its default configuration gaussian white noise is added to all pixels of the image.

The random number generators provided to this class have to be continuous distributions, which means they generate floating point values. For the fixed point types, like int and ubyte, the normalization constants provided by cvr::typeInfo<T>::suggestedNorm() are multiplied to the values generated by the distribution, prior to the addition to the actual value of each matrix element.

You can indicate with the parameter coverage how much of the pixels do you want to be altered with noise, where the specific pixels chosen for modification (if less than 100%) are also taken randomly.

The following example uses noise to generate a sample image with different degrees of noise.

   #include "cvrViewer2D.h"
   #include "cvrMath.h"
   #include "cvrNoise.h"
   #include "cvrNormalDistribution.h"

   // ...

   channel chnl,chnl2,nchnl;
   viewer2D view("Original");

   noise noiser;

   // create a black background with a grey square on which a
   // white square lies.
   chnl.assign(100,100,0.0f);   // 100x100 with black
   chnl.fill(0.5f,20,20,79,79); // leave a border of 20
   chnl.fill(1.0f,40,40,59,59); // the white square

   // a large canvas to place 5x5 images like the channel just created
   chnl2.allocate(chnl.rows()*5,chnl.columns()*5);

   // use a normal distribution
   normalDistribution nd;
   normalDistribution::parameters ndp;

   for (int y=0;y<5;++y) {
     for (int x=0;x<5;++x) {
       ndp.sigma = (y*5 + x)/50.0;   // set the standard deviation
       nd.setParameters(ndp);        // into the normal distribution
       noiser.setNoiseGenerator(nd); // which has to be used

       noiser.apply(chnl,nchnl);     // add the noise

       // place the new noisy patch on the canvas
       chnl2.fill(nchnl,y*chnl.rows(),x*chnl.columns());
     }
   }

   // show the canvas
   view.show(chnl2);

See also:: noise::parameters.

Constructor & Destructor Documentation

cvr::noise::noise ( )

Default constructor.

cvr::noise::noise ( const parameters & par )

Construct a functor using the given parameters.

cvr::noise::noise ( const noise & other )

Copy constructor.

Parameters:

other

the object to be copied

virtual cvr::noise::~noise ( ) [virtual]

Destructor.