Camera pose estimation

contains functions for estimating camera pose from different correspondences. More...

Functions
int	CVD::three_point_pose (const std::array< TooN::Vector< 3 >, 3 > &x, const std::array< TooN::Vector< 2 >, 3 > &z, std::vector< TooN::SE3<>> &poses)
	The function for pose estimation from three 2D - 3D point correspondences. More...

Detailed Description

contains functions for estimating camera pose from different correspondences.

Function Documentation

three_point_pose()

int CVD::three_point_pose	(	const std::array< TooN::Vector< 3 >, 3 > &	x,
		const std::array< TooN::Vector< 2 >, 3 > &	z,
		std::vector< TooN::SE3<>> &	poses
	)

The function for pose estimation from three 2D - 3D point correspondences.

It implements the algorithm given by the Fischer and Bolles RANSAC paper, 1980. This function assumes that the three points are in general position (not collinear). Input is an array of 3D cartesian positions and an array of 2D vectors that are the perspective projections of the points. Ouput is up to four poses satisfying the input with positive depths (points in front of the camera).

Parameters

[in]	x	an array containing at least 3 points
[in]	z	an array containing the perspective projections of the points given by x in the current pose
[out]	poses	the vector onto which any valid poses are appended

Returns

the number of poses appended to the vector

It implements the algorithm given by the Fischer and Bolles RANSAC paper, 1980. This function assumes that the three points are in general position (not collinear). Input is an array of 3D cartesian positions and an array of 3D vectors that are rays corresponding to (unprojected) 2D observations of the 3D points. Ouput is up to four poses satisfying the input with positive depths (points in front of the camera).

Parameters

[in]	x	an array containing at least 3 points
[in]	z	an array containing the rays corresponding to image points cast by x in the current pose
[out]	poses	the vector onto which any valid poses are appended

Returns

the number of poses appended to the vector