Image Analysis And Image Recognition at Universität Weimar | Flashcards & Summaries

Lernmaterialien für Image Analysis and Image Recognition an der Universität Weimar

Greife auf kostenlose Karteikarten, Zusammenfassungen, Übungsaufgaben und Altklausuren für deinen Image Analysis and Image Recognition Kurs an der Universität Weimar zu.

TESTE DEIN WISSEN

Given a sequence of points  lying on the object boundary briefly describe how to obtain its Fourier descriptors and why is the Fourier descriptor capable of dealing with noisy boundaries?

Lösung anzeigen
TESTE DEIN WISSEN
  • Given a set of 2d-coordinates (points) on a boundary
  • start from an arbitrary point
  • traverse the boundary and obtain a sequence of compley numbers S
  • compute DFT (discrete fourier transform) of F = DFT(S)
  • the coefficients in F are Fourier descriptors
  • usually the high frequency component in F have a small magnitute and will be removed
  • by doing this the small fluctuations or high-variation parts in the boundary will be smoothed, achieving the noise reduction effect
Lösung ausblenden
TESTE DEIN WISSEN

Definition Extraction


Lösung anzeigen
TESTE DEIN WISSEN

1. First erode  A by B 

2. subtact  result form original A


ß(A) = A - (A⊖B)


Lösung ausblenden
TESTE DEIN WISSEN

Basic Idea of SIFT

Lösung anzeigen
TESTE DEIN WISSEN

SIFT (Scale invariant Feature Transform)


Description computation:

  • Divide patch into 4x4 subpatches: 16 cells
  • compute  histogram of gradient orientations ( 8 reference angles) for all pixels inside each sub-patch
  • REsulting descriptor: 4x4x8 = 128 dimensions
  • Can handle changes in viewpoint up to 60° out-ot-plane rotation
  • can handle significant changes in illumination
  • fast and efficient - can run in real-time
Lösung ausblenden
TESTE DEIN WISSEN

Standard Morphological operators

Lösung anzeigen
TESTE DEIN WISSEN
  • Dialation
  • Erosion


  • Vereinigung
  • Schnittmengenbildung  
  • Mengendifferenzbildung
Lösung ausblenden
TESTE DEIN WISSEN

Bayer Filter

Lösung anzeigen
TESTE DEIN WISSEN
  • Fotosensor, ähnlich einem Schachbrett
  • mit Farbfilter überzogen
  • 50% grün, 25% rot, 25% blau
  • menschliche Auge reagiert auf grün empfindlicher als auf andere Farben
Lösung ausblenden
TESTE DEIN WISSEN

Definition Demosaicing

Lösung anzeigen
TESTE DEIN WISSEN

Rekonstruktion einer farbigen Rastergrafik aus den Helligkeitswerten eines mit Mosaik-Farbfiltern überlagerten Bildsensoren


Estimating missing components from values

Lösung ausblenden
TESTE DEIN WISSEN

Name the main  stages of the Canny  edge detector.

Lösung anzeigen
TESTE DEIN WISSEN

1:Gaussian Filter
2:Finding intensity gradient of image
3:Non-maximum suppression: – Thin multi-pixel wide “ridges” down to single pixel width
4:Hysteresis thresholding:
– Define two thresholds: low and high
– Use the high threshold to start edge curves and the low threshold to continue them.

Lösung ausblenden
TESTE DEIN WISSEN

What are  the basic steps for filtering in the frequency domain?

Lösung anzeigen
TESTE DEIN WISSEN

1:Fourier Transform image and filter
padding if required.

2:Multiply element wise in frequency domain

3:Inverse Transform

Lösung ausblenden
TESTE DEIN WISSEN

Pros/Cons  Supervised Classification

Lösung anzeigen
TESTE DEIN WISSEN

Advantages:

  • analyst has full control of the process
  • processing is tied to specific areas of knonw identify
  • analyst not faced with the problem of matching categories on dinal map with field inormation
  • operator can detect errors, often able the remedy them


Disadvantages:

  • Signatures are forced, because training classes are based  on  filed identification and not on spectral properties
  • training data selected by the analyst may not be representive
  • the preparation of training data is time-consuming and costly
  • it is not possible to recognise and represent special or unique categories which may be present in the image but are not represented in the training data
Lösung ausblenden
TESTE DEIN WISSEN

What is the meaning of the cluster centers?

Lösung anzeigen
TESTE DEIN WISSEN

The meaning of cluster centers depend on the methos we used to classify them.


If we used k-means classification these are the mean points of each class from which the other points have  the smallest distance.


If we used mean-shift classification  the center is the local maxima of density.

Lösung ausblenden
TESTE DEIN WISSEN

Definition image binarization

Lösung anzeigen
TESTE DEIN WISSEN
  • Image binarization is the process of taking a grayscale image and converting it to black-and-white.
  • Essentially reducing the infomration contained within the image from 256 shades of gray to 2: black and white, binary image
Lösung ausblenden
TESTE DEIN WISSEN

Erosion

Lösung anzeigen
TESTE DEIN WISSEN

shrinking image

Lösung ausblenden
  • 3387 Karteikarten
  • 144 Studierende
  • 9 Lernmaterialien

Beispielhafte Karteikarten für deinen Image Analysis and Image Recognition Kurs an der Universität Weimar - von Kommilitonen auf StudySmarter erstellt!

Q:

Given a sequence of points  lying on the object boundary briefly describe how to obtain its Fourier descriptors and why is the Fourier descriptor capable of dealing with noisy boundaries?

A:
  • Given a set of 2d-coordinates (points) on a boundary
  • start from an arbitrary point
  • traverse the boundary and obtain a sequence of compley numbers S
  • compute DFT (discrete fourier transform) of F = DFT(S)
  • the coefficients in F are Fourier descriptors
  • usually the high frequency component in F have a small magnitute and will be removed
  • by doing this the small fluctuations or high-variation parts in the boundary will be smoothed, achieving the noise reduction effect
Q:

Definition Extraction


A:

1. First erode  A by B 

2. subtact  result form original A


ß(A) = A - (A⊖B)


Q:

Basic Idea of SIFT

A:

SIFT (Scale invariant Feature Transform)


Description computation:

  • Divide patch into 4x4 subpatches: 16 cells
  • compute  histogram of gradient orientations ( 8 reference angles) for all pixels inside each sub-patch
  • REsulting descriptor: 4x4x8 = 128 dimensions
  • Can handle changes in viewpoint up to 60° out-ot-plane rotation
  • can handle significant changes in illumination
  • fast and efficient - can run in real-time
Q:

Standard Morphological operators

A:
  • Dialation
  • Erosion


  • Vereinigung
  • Schnittmengenbildung  
  • Mengendifferenzbildung
Q:

Bayer Filter

A:
  • Fotosensor, ähnlich einem Schachbrett
  • mit Farbfilter überzogen
  • 50% grün, 25% rot, 25% blau
  • menschliche Auge reagiert auf grün empfindlicher als auf andere Farben
Mehr Karteikarten anzeigen
Q:

Definition Demosaicing

A:

Rekonstruktion einer farbigen Rastergrafik aus den Helligkeitswerten eines mit Mosaik-Farbfiltern überlagerten Bildsensoren


Estimating missing components from values

Q:

Name the main  stages of the Canny  edge detector.

A:

1:Gaussian Filter
2:Finding intensity gradient of image
3:Non-maximum suppression: – Thin multi-pixel wide “ridges” down to single pixel width
4:Hysteresis thresholding:
– Define two thresholds: low and high
– Use the high threshold to start edge curves and the low threshold to continue them.

Q:

What are  the basic steps for filtering in the frequency domain?

A:

1:Fourier Transform image and filter
padding if required.

2:Multiply element wise in frequency domain

3:Inverse Transform

Q:

Pros/Cons  Supervised Classification

A:

Advantages:

  • analyst has full control of the process
  • processing is tied to specific areas of knonw identify
  • analyst not faced with the problem of matching categories on dinal map with field inormation
  • operator can detect errors, often able the remedy them


Disadvantages:

  • Signatures are forced, because training classes are based  on  filed identification and not on spectral properties
  • training data selected by the analyst may not be representive
  • the preparation of training data is time-consuming and costly
  • it is not possible to recognise and represent special or unique categories which may be present in the image but are not represented in the training data
Q:

What is the meaning of the cluster centers?

A:

The meaning of cluster centers depend on the methos we used to classify them.


If we used k-means classification these are the mean points of each class from which the other points have  the smallest distance.


If we used mean-shift classification  the center is the local maxima of density.

Q:

Definition image binarization

A:
  • Image binarization is the process of taking a grayscale image and converting it to black-and-white.
  • Essentially reducing the infomration contained within the image from 256 shades of gray to 2: black and white, binary image
Q:

Erosion

A:

shrinking image

Image Analysis and Image Recognition

Erstelle und finde Lernmaterialien auf StudySmarter.

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

Jetzt loslegen

Das sind die beliebtesten StudySmarter Kurse für deinen Studiengang Image Analysis and Image Recognition an der Universität Weimar

Für deinen Studiengang Image Analysis and Image Recognition 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 Image Analysis and Image Recognition Kurse im gesamten StudySmarter Universum

Action and Cognition

TU Kaiserslautern

Zum Kurs
Analysis I

Universität Stuttgart

Zum Kurs
Analysis Reußner

IU Internationale Hochschule

Zum Kurs
Brain and Cognition

University of Amsterdam

Zum Kurs

Die all-in-one Lernapp für Studierende

Greife auf Millionen geteilter Lernmaterialien der StudySmarter Community zu
Kostenlos anmelden Image Analysis and Image Recognition
Erstelle Karteikarten und Zusammenfassungen mit den StudySmarter Tools
Kostenlos loslegen Image Analysis and Image Recognition