Robust Estimation

Detailed Description

Functions
double	cv::sfm::fundamentalFromCorrespondences7PointRobust (InputArray x1, InputArray x2, double max_error, OutputArray F, OutputArray inliers, double outliers_probability=1e-2)
	Estimate robustly the fundamental matrix between two dataset of 2D point (image coords space).
double	cv::sfm::fundamentalFromCorrespondences8PointRobust (InputArray x1, InputArray x2, double max_error, OutputArray F, OutputArray inliers, double outliers_probability=1e-2)
	Estimate robustly the fundamental matrix between two dataset of 2D point (image coords space).

Function Documentation

fundamentalFromCorrespondences7PointRobust()

double cv::sfm::fundamentalFromCorrespondences7PointRobust	(	InputArray	x1,
		InputArray	x2,
		double	max_error,
		OutputArray	F,
		OutputArray	inliers,
		double	outliers_probability = 1e-2 )

#include <opencv2/sfm/robust.hpp>

Estimate robustly the fundamental matrix between two dataset of 2D point (image coords space).

Parameters

x1	Input 2xN Array of 2D points in view 1.
x2	Input 2xN Array of 2D points in view 2.
max_error	maximum error (in pixels).
F	Output 3x3 fundamental matrix such that \(x_2^T F x_1=0\).
inliers	Output 1xN vector that contains the indexes of the detected inliers.
outliers_probability	outliers probability (in ]0,1[). The number of iterations is controlled using the following equation: \(k = \frac{log(1-p)}{log(1.0 - w^n )}\) where \(k\), \(w\) and \(n\) are the number of iterations, the inliers ratio and minimun number of selected independent samples. The more this value is high, the less the function selects ramdom samples.

The fundamental solver relies on the 7 point solution. Returns the best error (in pixels), associated to the solution F.

fundamentalFromCorrespondences8PointRobust()

double cv::sfm::fundamentalFromCorrespondences8PointRobust	(	InputArray	x1,
		InputArray	x2,
		double	max_error,
		OutputArray	F,
		OutputArray	inliers,
		double	outliers_probability = 1e-2 )

#include <opencv2/sfm/robust.hpp>

Estimate robustly the fundamental matrix between two dataset of 2D point (image coords space).

Parameters

x1	Input 2xN Array of 2D points in view 1.
x2	Input 2xN Array of 2D points in view 2.
max_error	maximum error (in pixels).
F	Output 3x3 fundamental matrix such that \(x_2^T F x_1=0\).
inliers	Output 1xN vector that contains the indexes of the detected inliers.
outliers_probability	outliers probability (in ]0,1[). The number of iterations is controlled using the following equation: \(k = \frac{log(1-p)}{log(1.0 - w^n )}\) where \(k\), \(w\) and \(n\) are the number of iterations, the inliers ratio and minimun number of selected independent samples. The more this value is high, the less the function selects ramdom samples.

The fundamental solver relies on the 8 point solution. Returns the best error (in pixels), associated to the solution F.