maspack.geometry

Class CVReconstruction3d

java.lang.Object

maspack.geometry.CVReconstruction3d

public class CVReconstruction3d
extends java.lang.Object

Computer-Vision 3D reconstruction methods

Constructor Summary

Constructors

Constructor and Description
CVReconstruction3d()

Method Summary

Modifier and Type	Method and Description
static Matrix3d	computeCameraCalibration(double f, Point2d c) Determines the camera's intrinsic calibration matrix
static Matrix3x4	computeCameraMatrix(double f, Point2d c, RigidTransform3d T) Computes a camera matrix
static Matrix3x4	computeCameraMatrix(Matrix3d K, RigidTransform3d T) Transforms an intrinsic camera calibration matrix to a full camera matrix P
static Matrix3d	computeEssentialMatrix(Matrix3d F, Matrix3d K1, Matrix3d K2) Computes the essential matrix from intrinsic camera calibration and fundamental matrix
static Matrix3d	computeEssentialMatrix(RigidTransform3d T1, RigidTransform3d T2) Computes the essential matrix from extrinsic camera calibration
static Matrix3d	computeFundamentalMatrix(Matrix3x4 P1, Matrix3x4 P2) Compute fundamental matrix from two general camera matrices
static RigidTransform3d[]	computeRelativeTransform(Matrix3d F, Matrix3d K1, Matrix3d K2) Computes the relative transform between two cameras by decomposing the essential matrix.
static Vector4d	homogeneousReconstruction(Matrix3x4 P1, Matrix3x4 P2, Point2d p1, Point2d p2)
static RigidTransform3d	linearEightPointAlgorithm(java.util.Collection<? extends Point2d> p1, java.util.Collection<? extends Point2d> p2, Matrix3d K1, Matrix3d K2) Performs the eight-point algorithm for determining the essential matrix and recovering the extrinsic camera transform.
static Matrix3d	linearNormalizedEightPointAlgorithm(java.util.Collection<? extends Point2d> p1, java.util.Collection<? extends Point2d> p2) Compute fundamental matrix from a set of corresponding 2D points

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

CVReconstruction3d

public CVReconstruction3d()

Method Detail

computeCameraCalibration

public static Matrix3d computeCameraCalibration(double f,
                                                Point2d c)

Determines the camera's intrinsic calibration matrix

Parameters:

f - focal distance

c - principal point (center, optional - defaults to zero)

Returns:

intrinsic calibration matrix

computeCameraMatrix

public static Matrix3x4 computeCameraMatrix(Matrix3d K,
                                            RigidTransform3d T)

Transforms an intrinsic camera calibration matrix to a full camera matrix P

Parameters:

K - intrinsic camera calibration

T - extrinsic camera calibration

Returns:

full camera matrix

computeCameraMatrix

public static Matrix3x4 computeCameraMatrix(double f,
                                            Point2d c,
                                            RigidTransform3d T)

Computes a camera matrix

Parameters:

f - uniform focal distance

c - principal point (center, optional - defaults to zero)

T - camera-to-world transform (optional - defaults to identity)

Returns:

full camera matrix

computeFundamentalMatrix

public static Matrix3d computeFundamentalMatrix(Matrix3x4 P1,
                                                Matrix3x4 P2)

Compute fundamental matrix from two general camera matrices

Parameters:

P1 - camera matrix 1

P2 - camera matrix 2

Returns:

fundamental matrix

computeEssentialMatrix

public static Matrix3d computeEssentialMatrix(Matrix3d F,
                                              Matrix3d K1,
                                              Matrix3d K2)

Computes the essential matrix from intrinsic camera calibration and fundamental matrix

Parameters:

F - fundamental matrix between two cameras

K1 - camera 1 intrinsic calibration matrix

K2 - camera 2 intrinsic calibration matrix

Returns:

essential matrix

computeEssentialMatrix

public static Matrix3d computeEssentialMatrix(RigidTransform3d T1,
                                              RigidTransform3d T2)

Computes the essential matrix from extrinsic camera calibration

Parameters:

T1 - camera 1 world transform

T2 - camera 2 world transform

Returns:

essential matrix

computeRelativeTransform

public static RigidTransform3d[] computeRelativeTransform(Matrix3d F,
                                                          Matrix3d K1,
                                                          Matrix3d K2)

Computes the relative transform between two cameras by decomposing the essential matrix. See Zisserman and Hartley, 2004, pg 257-259.

Parameters:

F - fundamental matrix

K1 - camera 1 intrinsic calibration matrix

K2 - camera 2 intrinsic calibration matrix

Returns:

all four possible relative transforms

linearNormalizedEightPointAlgorithm

public static Matrix3d linearNormalizedEightPointAlgorithm(java.util.Collection<? extends Point2d> p1,
                                                           java.util.Collection<? extends Point2d> p2)

Compute fundamental matrix from a set of corresponding 2D points

Parameters:

p1 - set of points in camera 1

p2 - set of points in camera 2

Returns:

fundamental matrix

homogeneousReconstruction

public static Vector4d homogeneousReconstruction(Matrix3x4 P1,
                                                 Matrix3x4 P2,
                                                 Point2d p1,
                                                 Point2d p2)

linearEightPointAlgorithm

public static RigidTransform3d linearEightPointAlgorithm(java.util.Collection<? extends Point2d> p1,
                                                         java.util.Collection<? extends Point2d> p2,
                                                         Matrix3d K1,
                                                         Matrix3d K2)

Performs the eight-point algorithm for determining the essential matrix and recovering the extrinsic camera transform.

Parameters:

p1 - set of points in camera 1

p2 - set of points in camera 2

K1 - intrinsic camera calibration for camera 1

K2 - intrinsic camera calibration for camera 2

Returns:

extrinsic camera calibration matrix for camera 2 relative to camera 1