TRTK::RansacGenericFittingModel< T > Class Template Reference

This class implements the Ransac::Model interface for all fitting classes. More...

#include <RansacGenericFittingModel.hpp>

Inheritance diagram for TRTK::RansacGenericFittingModel< T >:

Collaboration diagram for TRTK::RansacGenericFittingModel< T >:

List of all members.

Public Member Functions
	RansacGenericFittingModel (Fit< T > &estimator)
	Constructs an instance of RansacGenericFittingModel.
void	compute ()
	Estimates the model parameters.
T	getDeviation (const Coordinate< T > &datum) const
	Returns the amount of how much a datum deviates from the model.
unsigned	getMinimumNumberOfItems () const
	Returns the minimum number of items required to compute the model.
T	getRMS () const
	Returns the root mean square error of the estimated regression model.
void	setData (const std::vector< Coordinate< T > > &data)
	Sets the sample data.
void	setEstimator (Fit< T > &estimator)
	Sets the fitting class instance.
virtual T	getDeviation (const DataType &datum) const =0
	Return the amount of how much a datum deviates from the model.
virtual T	getRMSE () const =0
	Return the root mean square error of the estimated regression model.
virtual void	setData (const std::vector< DataType > &data)=0
	Set the sample data. No data is copied but a reference is stored.

---

Detailed Description

template<class T>
class TRTK::RansacGenericFittingModel< T >

This class implements the Ransac::Model interface for all fitting classes.

Template Parameters:

T	Scalar type (must be a floating point).

Here is an example that shows how to use this class:

 #include <iostream>

 #include <TRTK/FitLine.hpp>
 #include <TRTK/RansacGenericFittingModel.hpp>
 #include <TRTK/Tools.hpp>

 using namespace std;
 using namespace TRTK;
 using namespace TRTK::Tools;

 int main()
 {
     // Construct some points lying on a line and add some noise and outliers.

     vector<Coordinate<double> > points;

     double slope = 0.7;
     double y_intercept = -3;

     for (int i = -10; i < 10; ++i)
     {
         // Noisy measurement.

         double x = i + randn(0.0, 0.1);
         double y = i * slope + y_intercept + randn(0.0, 0.1);

         Coordinate<double> point(x, y);

         points.push_back(point);
     }

     for (int i = 0; i < 5; ++i)
     {
         // Gros outliers.

         double x = rand(-10.0, 10.0);
         double y = rand(-10.0, 10.0);

         Coordinate<double> point(x, y);

         points.push_back(point);
     }

     // Estimate the line parameters using ordinary least sqares.

     FitLine<double> fitLine(points);

     fitLine.compute();

     cout << "Slope: " << fitLine.getSlope() << endl;
     cout << "Y-intercept: " << fitLine.getYIntercept() << endl;
     cout << "Direction Vector: " << fitLine.getDirectionVector() << endl;
     cout << "Distance from origin: " << fitLine.getDistanceFromOrigin() << endl;
     cout << "RMS: " << fitLine.getRMS() << endl << endl;

     // Estimate the line parameters using RANSAC.

     RansacGenericFittingModel<double> model(fitLine);

     Ransac<double> ransac;

     ransac.setModel(model);
     ransac.setData(points);
     ransac.setErrorTolerance(0.2);

     unsigned number_of_samples_used = ransac.compute();

     cout << "Slope: " << fitLine.getSlope() << endl;
     cout << "Y-intercept: " << fitLine.getYIntercept() << endl;
     cout << "Direction Vector: " << fitLine.getDirectionVector() << endl;
     cout << "Distance from origin: " << fitLine.getDistanceFromOrigin() << endl;
     cout << "Number of samples used: " << number_of_samples_used << endl;
     cout << "RMS: " << model.getRMS() << endl;

     return 0;
 }

Output:

 Slope: 0.824709
 Y-intercept: -2.97947
 Direction Vector: (-0.771483, -0.63625)
 Distance from origin: 2.29861
 RMS: 2.27904

 Slope: 0.691347
 Y-intercept: -3.00877
 Direction Vector: (-0.822562, -0.568676)
 Distance from origin: 2.4749
 Number of samples used: 19
 RMS: 0.0712661

For further information, please have a look at the documentation of the particular fitting class and the Ransac class, respectively.

See also:

Ransac, Ransac::Model, Fit

Author:

Christoph Haenisch

Version:

0.1.0

Date:

last changed on 2012-03-23

Definition at line 140 of file RansacGenericFittingModel.hpp.

---

Constructor & Destructor Documentation

template<class T >

TRTK::RansacGenericFittingModel< T >::RansacGenericFittingModel

(

Fit< T > &

estimator )

Constructs an instance of RansacGenericFittingModel.

Parameters:

[in]

model

A fitting class.

Definition at line 166 of file RansacGenericFittingModel.hpp.

---

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

• D:/Programmierung/C/TRTK/include/TRTK/RansacGenericFittingModel.hpp