 Photogrammetric Computer Vision an der Universität Weimar | Karteikarten & Zusammenfassungen

# Lernmaterialien für photogrammetric computer vision an der Universität Weimar

Greife auf kostenlose Karteikarten, Zusammenfassungen, Übungsaufgaben und Altklausuren für deinen photogrammetric computer vision Kurs an der Universität Weimar zu.

TESTE DEIN WISSEN

What are major tasks for a 3D object reconstruction?

Lösung anzeigen
TESTE DEIN WISSEN

- Perspective

-Motion

-Silhouette and color consistency

- Texture

Lösung ausblenden
TESTE DEIN WISSEN

What does one understand by DLT?

Lösung anzeigen
TESTE DEIN WISSEN

Direct Linear transformation

-originally limited to transformation betweend 3D space and 2D image coordinates

- now generalized on many problems of projective geometry

---> mathematical approach to problems in photogrammetry

For/Contra

+ direct unique solution (without iteration(initial value)

-minimization of algebraic error without geometrical/statistical meaning

Use DLT as initial value for non-linear optimization with meaningful geometrical cost function

Lösung ausblenden
TESTE DEIN WISSEN

What is a homogeneous equation system and what must be considered for a stable solution?

Lösung anzeigen
TESTE DEIN WISSEN

In general, the equation AX=B representing a system of equations is called homogeneous if B is the nx1 (column) vector of zeros. Otherwise, the equation is called nonhomogeneous.

Linear homogeneous equation systems

Ax = 0

A = Design matrix (wth obeservations(measurements))

x wanted solution vector

unique solution , if at least as many independent homogeneous equations are presented ass unknown parameters

Lösung ausblenden
TESTE DEIN WISSEN

Compare the methods for parameter estimation

a. linear (in-)homogeneous equation systems,

b. robust parameter estimation

c. non-linear optimization

concerning the requirements, complexity and accuracy.

Lösung anzeigen
TESTE DEIN WISSEN

Linear (in-)homogeneous equation systems:

Advantages: no initialization required, 1 iteration

Disadvantages: sensitive to impercise measurements and outliers, usually no minimal parametrization, constraints enforced afterwards

Robust parameter estimation:

- M-estimators

-RANSAC --> Robust fit of model to data set S, which contains outliers

non-linear optimization:

- often slower than  DLT

- requires initialization --> typically use linear solution or sample parameter space

- no guaranteed convergence, local minima

- stopping criterion required

Many optimization algorithms exist, minimize geometric error function by adjusting the 9 matrix elements in Homography

Lösung ausblenden
TESTE DEIN WISSEN

Specify for each case a procedure for the computation of the parameter estimation. (linear (in-)homogeneous equation systems, robust parameter estimation, non-linear optimization)

Lösung anzeigen
TESTE DEIN WISSEN

Linear (in-)homogeneous equation systems --> direct linear estimation

Robust parameter estimation --> RANSAC

Non-linear optimization --> Least  squares estimation

Lösung ausblenden
TESTE DEIN WISSEN

How many degrees of freedom have these spaces?

Lösung anzeigen
TESTE DEIN WISSEN

Euclidean space: 6 DoF --> rotation :3 + translation(x,y,z): 3

Metric space: 7 DoF --> rotation: 3 + translation(x,y): 3 + global scaling: 1

Affine space: 12 DoF --> Metric (rotation, translation, scaling): 7 + individual scaling: 3 + skew factor: 2

Projective space: 15 DoF --> Affine: 12 + persepctive projection --> 4x4 matrix with 15 DoF and last element is homogeneous

Lösung ausblenden
TESTE DEIN WISSEN

What are homogeneous coordinates and what are the advantages?

Lösung anzeigen
TESTE DEIN WISSEN

They are a system of coordinates used in  the projective  space.

Allow us to describe Eculidean objects in  the projective space

Allow us to describe points at infinity as well as calculate intersection of two parallels

- Representation of Euclidean by adding component and a free scaling to mathematical representation

- Redundant (überflüssige) representation of vectors and matrices

Lösung ausblenden
TESTE DEIN WISSEN

What does incidence mean?

Lösung anzeigen
TESTE DEIN WISSEN

Realtion between points, lines

Lösung ausblenden
TESTE DEIN WISSEN

What is a homography and which property has it?

Lösung anzeigen
TESTE DEIN WISSEN

Any two images of the same planar surface in space related by a homography:

- invertible mapping h from P^2 to P^2 exactly if three points x1, x2, x3 of a straight line are located again on a straight line after mapping h(x1), h(x2), h(x3)

h(x) = Hx

Lösung ausblenden
TESTE DEIN WISSEN

How many degrees of freedom does a

a. planar homography

b. spatial homography

c. projective transformation from space into a plane have?

Lösung anzeigen
TESTE DEIN WISSEN

planar homography: number of unknown parameter: n*(n+2) --> 8 DoF for plane

spatial homography: 3*(3+2)= 15 DoF

projective transformation from space into a plane: 3x4 projection matrix P (11 DoF)

Lösung ausblenden
TESTE DEIN WISSEN

How many points are needed in each case (planar, spatial, projectie transformation from space into a plane) for the computation?

Lösung anzeigen
TESTE DEIN WISSEN

planar homography: n*(n+2) --> n+2=2+2=4 --> at least 4 points

spatial homography: n*(n+2) --> n+2=3+2=5 --> at least 5 points

projective transformation: we need 11 equations/information, each object contains 3 components (x,y,z) , we lose depth, so we need 5 1/2 points --> gives us 6 points

Lösung ausblenden
TESTE DEIN WISSEN

What does the term photogrammetry mean?

Lösung anzeigen
TESTE DEIN WISSEN

- is the acquirement of geometric object information using photographic images

- the science and tecnology of deducing reliable spatial information from measurements on photographs

Lösung ausblenden • 3378 Karteikarten
• 141 Studierende
• 9 Lernmaterialien

## Beispielhafte Karteikarten für deinen photogrammetric computer vision Kurs an der Universität Weimar - von Kommilitonen auf StudySmarter erstellt!

Q:

What are major tasks for a 3D object reconstruction?

A:

- Perspective

-Motion

-Silhouette and color consistency

- Texture

Q:

What does one understand by DLT?

A:

Direct Linear transformation

-originally limited to transformation betweend 3D space and 2D image coordinates

- now generalized on many problems of projective geometry

---> mathematical approach to problems in photogrammetry

For/Contra

+ direct unique solution (without iteration(initial value)

-minimization of algebraic error without geometrical/statistical meaning

Use DLT as initial value for non-linear optimization with meaningful geometrical cost function

Q:

What is a homogeneous equation system and what must be considered for a stable solution?

A:

In general, the equation AX=B representing a system of equations is called homogeneous if B is the nx1 (column) vector of zeros. Otherwise, the equation is called nonhomogeneous.

Linear homogeneous equation systems

Ax = 0

A = Design matrix (wth obeservations(measurements))

x wanted solution vector

unique solution , if at least as many independent homogeneous equations are presented ass unknown parameters

Q:

Compare the methods for parameter estimation

a. linear (in-)homogeneous equation systems,

b. robust parameter estimation

c. non-linear optimization

concerning the requirements, complexity and accuracy.

A:

Linear (in-)homogeneous equation systems:

Advantages: no initialization required, 1 iteration

Disadvantages: sensitive to impercise measurements and outliers, usually no minimal parametrization, constraints enforced afterwards

Robust parameter estimation:

- M-estimators

-RANSAC --> Robust fit of model to data set S, which contains outliers

non-linear optimization:

- often slower than  DLT

- requires initialization --> typically use linear solution or sample parameter space

- no guaranteed convergence, local minima

- stopping criterion required

Many optimization algorithms exist, minimize geometric error function by adjusting the 9 matrix elements in Homography

Q:

Specify for each case a procedure for the computation of the parameter estimation. (linear (in-)homogeneous equation systems, robust parameter estimation, non-linear optimization)

A:

Linear (in-)homogeneous equation systems --> direct linear estimation

Robust parameter estimation --> RANSAC

Non-linear optimization --> Least  squares estimation

Q:

How many degrees of freedom have these spaces?

A:

Euclidean space: 6 DoF --> rotation :3 + translation(x,y,z): 3

Metric space: 7 DoF --> rotation: 3 + translation(x,y): 3 + global scaling: 1

Affine space: 12 DoF --> Metric (rotation, translation, scaling): 7 + individual scaling: 3 + skew factor: 2

Projective space: 15 DoF --> Affine: 12 + persepctive projection --> 4x4 matrix with 15 DoF and last element is homogeneous

Q:

What are homogeneous coordinates and what are the advantages?

A:

They are a system of coordinates used in  the projective  space.

Allow us to describe Eculidean objects in  the projective space

Allow us to describe points at infinity as well as calculate intersection of two parallels

- Representation of Euclidean by adding component and a free scaling to mathematical representation

- Redundant (überflüssige) representation of vectors and matrices

Q:

What does incidence mean?

A:

Realtion between points, lines

Q:

What is a homography and which property has it?

A:

Any two images of the same planar surface in space related by a homography:

- invertible mapping h from P^2 to P^2 exactly if three points x1, x2, x3 of a straight line are located again on a straight line after mapping h(x1), h(x2), h(x3)

h(x) = Hx

Q:

How many degrees of freedom does a

a. planar homography

b. spatial homography

c. projective transformation from space into a plane have?

A:

planar homography: number of unknown parameter: n*(n+2) --> 8 DoF for plane

spatial homography: 3*(3+2)= 15 DoF

projective transformation from space into a plane: 3x4 projection matrix P (11 DoF)

Q:

How many points are needed in each case (planar, spatial, projectie transformation from space into a plane) for the computation?

A:

planar homography: n*(n+2) --> n+2=2+2=4 --> at least 4 points

spatial homography: n*(n+2) --> n+2=3+2=5 --> at least 5 points

projective transformation: we need 11 equations/information, each object contains 3 components (x,y,z) , we lose depth, so we need 5 1/2 points --> gives us 6 points

Q:

What does the term photogrammetry mean?

A:

- is the acquirement of geometric object information using photographic images

- the science and tecnology of deducing reliable spatial information from measurements on photographs ### Erstelle und finde Lernmaterialien auf StudySmarter.

Greife kostenlos auf tausende geteilte Karteikarten, Zusammenfassungen, Altklausuren und mehr zu.

## Das sind die beliebtesten StudySmarter Kurse für deinen Studiengang photogrammetric computer vision an der Universität Weimar

Für deinen Studiengang photogrammetric computer vision an der Universität Weimar gibt es bereits viele Kurse, die von deinen Kommilitonen auf StudySmarter erstellt wurden. Karteikarten, Zusammenfassungen, Altklausuren, Übungsaufgaben und mehr warten auf dich!

## Das sind die beliebtesten photogrammetric computer vision Kurse im gesamten StudySmarter Universum

TU München

##### computer vision

Hochschule Karlsruhe

##### Computer Vision

Hochschule Karlsruhe

##### Photogrammetrie

Hochschule Karlsruhe

## Die all-in-one Lernapp für Studierende

##### Greife auf Millionen geteilter Lernmaterialien der StudySmarter Community zu ##### Erstelle Karteikarten und Zusammenfassungen mit den StudySmarter Tools 