Aim: Implements a functor to detect feature points from normal tensor voting strategy. More...

#include <DGtal/geometry/surfaces/estimation/estimationFunctors/TensorVotingFeatureExtraction.h>

Public Types
typedef TSurfel	Surfel
typedef TEmbedder	SCellEmbedder
typedef SCellEmbedder::RealPoint	RealPoint
typedef double	Quantity

Public Member Functions
	TensorVotingFeatureExtraction (ConstAlias< SCellEmbedder > anEmbedder, const double h)
void	pushSurfel (const Surfel &aSurf, const double aDistance)
Quantity	eval ()
void	reset ()

Private Attributes
bool	myFirstSurfel
	First surfel flag.
const SCellEmbedder *	myEmbedder
	Alias of the geometrical embedder.
double	myH
RealPoint	myReceiver
	Receiver (embedding of the starting surfel)
SimpleMatrix< double, 3, 3 >	myAccum
	Covariance accumulator.
SimpleMatrix< double, 3, 3 >	myId
SimpleMatrix< double, 3, 3 >	myVote
double	myArea
	Convolution kernel area.

Detailed Description

template<typename TSurfel, typename TEmbedder>
class DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >

Aim: Implements a functor to detect feature points from normal tensor voting strategy.

Description of template class 'TensorVotingFeatureExtraction'

More precisely, the functor accumulates tensor votes \(I - vv^T/\|vv^T\|\) for each surfel added during the scan ( \(v\) being the vector from the center of the neighborhood to the added surfel). Then, the eval() method returns the ratio \(\frac{\lambda_1+\lambda_2}{\lambda_3}\) of the eigenvalues ( \( \lambda_1\leq\lambda_2\leq\lambda_3\)) of the accumulated tensor votes.

(see [98])

model of concepts::CLocalEstimatorFromSurfelFunctor

Template Parameters

TSurfel	type of surfels
TEmbedder	type of functors which embed surfel to \( \mathbb{R}^3\)

Definition at line 81 of file TensorVotingFeatureExtraction.h.

Member Typedef Documentation

◆ Quantity

template<typename TSurfel, typename TEmbedder>

typedef double DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::Quantity

Definition at line 88 of file TensorVotingFeatureExtraction.h.

◆ RealPoint

template<typename TSurfel, typename TEmbedder>

typedef SCellEmbedder::RealPoint DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::RealPoint

Definition at line 87 of file TensorVotingFeatureExtraction.h.

◆ SCellEmbedder

template<typename TSurfel, typename TEmbedder>

typedef TEmbedder DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::SCellEmbedder

Definition at line 86 of file TensorVotingFeatureExtraction.h.

◆ Surfel

template<typename TSurfel, typename TEmbedder>

typedef TSurfel DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::Surfel

Definition at line 85 of file TensorVotingFeatureExtraction.h.

Constructor & Destructor Documentation

◆ TensorVotingFeatureExtraction()

template<typename TSurfel, typename TEmbedder>

DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::TensorVotingFeatureExtraction	(	ConstAlias< SCellEmbedder >	anEmbedder,
		const double	h )

inline

Constructor.

Parameters

anEmbedder	embedder to map surfel to R^n.
h	grid step

Definition at line 96 of file TensorVotingFeatureExtraction.h.

                                                   :
        myEmbedder(&anEmbedder), myH(h)
      {
        myId.identity();
        myArea = 0.0;
        myFirstSurfel = true;
        myAccum.constant(0.0);
      }

References myAccum, myArea, myEmbedder, myFirstSurfel, myH, and myId.

Member Function Documentation

◆ eval()

template<typename TSurfel, typename TEmbedder>

Quantity DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::eval ( )

inline

Evaluate the feature score.

I.e. (l_1+l_2)/l_3 of the tensor eigenvalues (l_1<l_2<l_3).

Returns: the feature score

Definition at line 154 of file TensorVotingFeatureExtraction.h.

      {
        SimpleMatrix<double, 3, 3> eigenvectors;
        RealPoint eigenvalues;
 
        myAccum /= myArea;
        EigenDecomposition<3, double>::getEigenDecomposition( myAccum, eigenvectors, eigenvalues);
  
#ifdef DEBUG
        for( Dimension i_dim = 1; i_dim < 3; ++i_dim )
            ASSERT ( std::abs(eigenvalues[i_dim - 1]) <= std::abs(eigenvalues[i_dim]) );
#endif
      
        return ((eigenvalues[0] + eigenvalues[1])/(eigenvalues[2]));
      }

References DGtal::EigenDecomposition< TN, TComponent, TMatrix >::getEigenDecomposition(), myAccum, and myArea.

◆ pushSurfel()

template<typename TSurfel, typename TEmbedder>

void DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::pushSurfel	(	const Surfel &	aSurf,
		const double	aDistance )

inline

Add the geometrical embedding of a surfel to the point list and update the tensor voting.

Parameters

aSurf	a surfel to add
aDistance	distance of aSurf to the neighborhood boundary

Definition at line 113 of file TensorVotingFeatureExtraction.h.

      {
        if (myFirstSurfel)
          {
            myReceiver =  myEmbedder->operator()(aSurf);
            myFirstSurfel = false;
          }
        else
          {
            myArea+= aDistance;
            const RealPoint p = myEmbedder->operator()(aSurf);
            const RealPoint v = p - myReceiver;
          
            double maxcol = 0.0;
            double matnorm = 0.0;
          
            // I - vv^t/||vv^t||
            for(DGtal::Dimension i= 0; i <3; i++)
              {
                maxcol = 0.0;
                for(DGtal::Dimension j=0; j < 3; j++)
                  {
                    myVote.setComponent(i,j, v(i)*v(j));
                    if (std::abs(v(i)*v(j)) > maxcol)
                      maxcol = std::abs(v(i)*v(j));
                  }
                matnorm += maxcol;
              }
        
            myAccum += (myId - myVote/matnorm)*aDistance;
          }
      }

References myAccum, myArea, myEmbedder, myFirstSurfel, myId, myReceiver, and myVote.

◆ reset()

template<typename TSurfel, typename TEmbedder>

void DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::reset ( )

inline

Reset the point list.

Definition at line 174 of file TensorVotingFeatureExtraction.h.

      {
        myArea = 0.0;
        myFirstSurfel = true;
        myAccum.constant(0.0);
      }

References myAccum, myArea, and myFirstSurfel.

Field Documentation

◆ myAccum

template<typename TSurfel, typename TEmbedder>

SimpleMatrix<double, 3, 3> DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::myAccum

private

Covariance accumulator.

Definition at line 197 of file TensorVotingFeatureExtraction.h.

Referenced by eval(), pushSurfel(), reset(), and TensorVotingFeatureExtraction().

◆ myArea

template<typename TSurfel, typename TEmbedder>

double DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::myArea

private

Convolution kernel area.

Definition at line 200 of file TensorVotingFeatureExtraction.h.

Referenced by eval(), pushSurfel(), reset(), and TensorVotingFeatureExtraction().

◆ myEmbedder

template<typename TSurfel, typename TEmbedder>

const SCellEmbedder* DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::myEmbedder

private

Alias of the geometrical embedder.

Definition at line 188 of file TensorVotingFeatureExtraction.h.

Referenced by pushSurfel(), and TensorVotingFeatureExtraction().

◆ myFirstSurfel

template<typename TSurfel, typename TEmbedder>

bool DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::myFirstSurfel

private

First surfel flag.

Definition at line 185 of file TensorVotingFeatureExtraction.h.

Referenced by pushSurfel(), reset(), and TensorVotingFeatureExtraction().

◆ myH

template<typename TSurfel, typename TEmbedder>

double DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::myH

private

Definition at line 191 of file TensorVotingFeatureExtraction.h.

Referenced by TensorVotingFeatureExtraction().

◆ myId

template<typename TSurfel, typename TEmbedder>

SimpleMatrix<double, 3, 3> DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::myId

private

Definition at line 197 of file TensorVotingFeatureExtraction.h.

Referenced by pushSurfel(), and TensorVotingFeatureExtraction().

◆ myReceiver

template<typename TSurfel, typename TEmbedder>

RealPoint DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::myReceiver

private

Receiver (embedding of the starting surfel)

Definition at line 194 of file TensorVotingFeatureExtraction.h.

Referenced by pushSurfel().

◆ myVote

template<typename TSurfel, typename TEmbedder>

SimpleMatrix<double, 3, 3> DGtal::functors::TensorVotingFeatureExtraction< TSurfel, TEmbedder >::myVote

private

Definition at line 197 of file TensorVotingFeatureExtraction.h.

Referenced by pushSurfel().

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

TensorVotingFeatureExtraction.h

Public Types

Public Member Functions

Private Attributes

Detailed Description

Member Typedef Documentation

◆ Quantity

◆ RealPoint

◆ SCellEmbedder

◆ Surfel

Constructor & Destructor Documentation

◆ TensorVotingFeatureExtraction()

Member Function Documentation

◆ eval()

◆ pushSurfel()

◆ reset()

Field Documentation

◆ myAccum

◆ myArea

◆ myEmbedder

◆ myFirstSurfel

◆ myH

◆ myId

◆ myReceiver

◆ myVote