ldvSBA.h File Reference

---

Detailed Description

Several function recovering homographies based on pairs of images or even optimizing over all images using block bundle adjustment.

Author:

Copyright (C) 2008 Vincent Rabaud <vrabaud@cs.ucsd.edu>

Version:

1.2-20080501

#include <vector>
#include "sba.h"

Include dependency graph for ldvSBA.h:

This graph shows which files directly or indirectly include this file:

Classes
struct	Distortion
	Holds the distortion informations. More...
Functions
int	getFeatureGlobalId (feature *feat)
	Extracts the id of 2D feature corresponding to a feature.
void	computePairwiseHomography (unsigned int imgInd1, unsigned int imgInd2, vector< ImgInfo > &vImgInfo, Match &match, AllFeatureGlobal &allFeatureGlobal)
	Compute a possible homography between two sets of features.
void	computeAllPairwiseHomography (vector< ImgInfo > &vImgInfo, Match &match, AllFeatureGlobal &allFeatureGlobal)
	Given images with matching SIFT features, compute possible homographies but only keep the ones with the most inliers.
void	computeGlobalHomography (vector< ImgInfo > &vImgInfo, unsigned int imgInd, Match &match, AllFeatureGlobal &allFeatureGlobal)
	Compute the homographies for each image to minimize a global matching error.
Variables
float	SIFT_RANSAC_THRES
	tolerance (in pixels) to which points ae considered as inliers in RANSAC
const double	SBA_TERMINATION_THRESH = 1E-20
	Defines when the SBA optimization has to be terminated.

---

Function Documentation

int getFeatureGlobalId

(

feature *

feat

)

Extracts the id of 2D feature corresponding to a feature.

Parameters:

feat

Pointer to a feature.

Returns:

The id of the 2D feature corresponding to the feature.

void computePairwiseHomography	(	unsigned int	imgInd1,
		unsigned int	imgInd2,
		vector< ImgInfo > &	vImgInfo,
		Match &	match,
		AllFeatureGlobal &	allFeatureGlobal
	)

Compute a possible homography between two sets of features.

Parameters:

	imgInd1	index of the first image
	imgInd2	index of the second image
	vImgInfo	vector containing information about the images
	match	contains all the information concerning matches
	allFeatureGlobal	all the FeatureGlobal

Returns:

the number of inliers

References AllFeatureGlobal::addNewPair(), AllFeatureGlobal::addSingle(), displayMatrix(), getFeatInd(), feature2::getFeatureGlobalInd(), getImgInd(), feature2::hasFeatGlob(), AllFeatureGlobal::merge(), Match::setComparedH(), Match::setH(), and SIFT_RANSAC_THRES.

Referenced by computeAllPairwiseHomography().

Here is the call graph for this function:

Here is the caller graph for this function:

void computeAllPairwiseHomography	(	vector< ImgInfo > &	vImgInfo,
		Match &	match,
		AllFeatureGlobal &	allFeatureGlobal
	)

Given images with matching SIFT features, compute possible homographies but only keep the ones with the most inliers.

Parameters:

	vImgInfo	vector containing information about the images
	match	contains all the information concerning matches
	allFeatureGlobal	all the FeatureGlobal

References computePairwiseHomography(), Match::getComparedH(), Match::getH(), and Match::getNMatch().

Here is the call graph for this function:

void computeGlobalHomography	(	vector< ImgInfo > &	vImgInfo,
		unsigned int	imgInd,
		Match &	match,
		AllFeatureGlobal &	allFeatureGlobal
	)

Compute the homographies for each image to minimize a global matching error.

Parameters:

	vImgInfo	vector containing information about the images
	imgInd	number of images to consider in vImgInfo
	match	contains all the information concerning matches
	allFeatureGlobal	vector containing the already existing 2D features

References computeCornerExtrema(), displayMatrix(), getImgInd(), ImgInfo::height, Distortion::kap1, Distortion::kap1Incr, Distortion::kap1Max, Distortion::kap1Min, SBA_TERMINATION_THRESH, ImgInfo::setH(), AllFeatureGlobal::size(), and ImgInfo::width.

Here is the call graph for this function: