Lecture 7&8 Flashcards

1
Q

Usage of Gabor Filter

A

Used to model the response of simple V1 cells

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Gabor filters

A

detecting parts of a signal with a specific frequency and orientation in a local region

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Gaussian envelope

A

localized in space and in the Fourier Transform

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

what are texture segmentation

A

surfaces with repeating patterns and structure

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What is the relation of different parameters to the Gabor filters?

A

Determine the shape and orientation of the field

𝜃 : orientation of the filter
𝛾 : aspect ratio (between the x and y axes)
σ: std of Gaussian envelope
ϕ: phase of the sine wave
λ: wavelength of sine grating

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Does the phase parameter affects the Magnitude plot

A

no, the phase does not affect the magnitude plots

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Gabor Filter bank

A

Collection of all orientations and spacial frequencies in an image

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Magnitude Plot of a Gabor Filter Bank

A

high spatial frequencies are closer to the DC point
low spatial frequencies are far away from the DC point

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Power Spectrum

A

magnitude plot of spatial frequencies

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

the Fourier amplitude of spatial frequencies is proportional to

A

1/f

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

proportion for the power spectrum

A

1/f^2

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Fall-off feature of a power spectrum

A

Power spectrum of natural images have 1/f^2
higher spatial freqs have lower power spectral density
lower spatial freqs have higher power spectral density

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Principal Component Analysis (PCA)

A

technique to compute the orthogonal directions of maximum possible variance in data

dimensionality reduction

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

How can we localize a receptive field of an image

A

by applying a gaussian mask

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Why is PCA mandatory

A

because images are redundat by removing the redundancy we can focus on the relevant information

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

How Information theory applies to the Visual System/ Computer vision

A

If we want to minimize the redundancy of information channel and maximize the information capacity we need to maximize the entropy

17
Q

global FT vs local FT

A

global FT: perform a FT to the whole image
local FT: perform a FT to the small image patches