mimas::warp< Tensor, T > Class Template Reference
[Image warping]

Image to image transform. More...

#include <warp.h>

Inheritance diagram for mimas::warp< Tensor, T >:

[legend]Collaboration diagram for mimas::warp< Tensor, T >:

[legend]List of all members.

Public Types
typedef boost::numeric::ublas::vector< double >	Vector
Public Member Functions
	warp (const Tensor &aT, T aBackground)
	Constructor.
mimas::image< T >	operator() (const mimas::image< T > &in) const
	Warp image.
Protected Attributes
Tensor	t
	The tensor.
T	background
	Background value for undefined pixels.

---

Detailed Description

template<class Tensor, typename T>
class mimas::warp< Tensor, T >

Image to image transform.

The image is warped using a 2D-tensor : , where is the middle of the image.

You need to define your own tensor function as function object:

// Lenscorrection
// Bala Amavasai (bala@amavasai.org)
// Jan Wedekind (jan at wedesoft.de)
// Mon Aug 23 14:37:05 UTC 2004

#include <boost/numeric/ublas/vector.hpp>
#include <functional>
#include <iostream>
#include <fstream>
#include "image.h"
#include "image_fileinput.h"
#include "image_mesaoutput.h"
#include "vectorfield.h"
#include "warp.h"

using namespace mimas;
using namespace std;

typedef boost::numeric::ublas::vector< double > Vector;

class LensCorrection: public std::unary_function< Vector, Vector >
{
 public:
  LensCorrection( double _pixelSizeX, double _pixelSizeY,
                  double _focalLength, double _curvature ):
    factorX( _pixelSizeX / _focalLength ),
    factorY( _pixelSizeY / _focalLength ),
    curvature( _curvature ) {}
  Vector operator()( const Vector &v ) const {
    double
      x = v( 0 ) * factorX,
      y = v( 1 ) * factorY,
      r = sqrt( x * x + y * y );
    double factor = ( 1.0 - curvature * r ) / ( 1.0 - curvature );
    Vector retVal( 2 );
    retVal( 0 ) = x * ( factor - 1.0 ) / factorX;
    retVal( 1 ) = y * ( factor - 1.0 ) / factorY;
    return retVal;    
  };
 protected:
  double factorX;
  double factorY;
  double curvature;
};

int main( int argc, char *argv[] )
{
  int retVal;
  try {
    MMERROR( argc == 3, mimasexception, ,
             "Syntax: " << argv[0] << " <image file> <correction>" );
    x11_display display;
    image_mesaoutput< unsigned char > window( &display );
    ifstream file( argv[1], ios::binary );
    image< int > inImage;
    MMERROR( file >> inImage, mimasexception, ,
             "Error reading from file \"" << argv[1] << "\"." );
    window << inImage << mimas::pause();
    vectorfield< LensCorrection > vectorField
      ( inImage.getWidth(), inImage.getHeight(),
        LensCorrection( 1.0 / inImage.getHeight(),
                        1.0 / inImage.getHeight(),
                        0.5, atof( argv[2] ) ) );

    // Instantiate warp-object.
    warp< vectorfield< LensCorrection >, int > w( vectorField, 0 );
    // Apply warp-object.
    image< int > outImage( w( inImage ) );
    
    // Display result.
    window << outImage << mimas::pause();
    retVal = 0;
  } catch ( exception &e ) {
    cerr << e.what() << endl;
    retVal = 1;
  };
  return retVal;
}

Todo:

Maybe the vectors passed to the tensor should be scaled, so that the tensor-specification will be independent of the image resolution.

Todo:

Provide functionality for using interpolation.

Author:

Jan Wedekind (jan at wedesoft.de)

Date:

Wed Jul 07 14:13:00 UTC 2004

See also:

vectorfield

Definition at line 28 of file warp.h.

---

Member Typedef Documentation

template<class Tensor, typename T>

typedef boost::numeric::ublas::vector< double > mimas::warp< Tensor, T >::Vector

Definition at line 31 of file warp.h.

---

Constructor & Destructor Documentation

template<class Tensor, typename T>

mimas::warp< Tensor, T >::warp	(	const Tensor &	aT,
		T	aBackground
	)			[inline]

Constructor.

Parameters:

	aT	tensor to apply to images.
	aBackground	Background value to use for undefined pixels.

Definition at line 35 of file warp.h.

---

Member Function Documentation

template<class Tensor, typename T>

mimas::image< T > mimas::warp< Tensor, T >::operator()

(

const mimas::image< T > &

in

)

const

Warp image.

Parameters:

in

Input image.

Returns:

Output image.

---

Member Data Documentation

template<class Tensor, typename T>

Tensor mimas::warp< Tensor, T >::t [protected]

The tensor.

Definition at line 42 of file warp.h.

template<class Tensor, typename T>

T mimas::warp< Tensor, T >::background [protected]

Background value for undefined pixels.

Definition at line 44 of file warp.h.

---

The documentation for this class was generated from the following file:

• warp.h

---