Filtering

Question 1

Sampling in 2D

Answer 1

Takes a function and returns a matrix

Question 2

Resolution

Answer 2

DPI of an image, spatial pixel density

Question 3

Quantization

Answer 3

8 bit-256 levels
4bit - 16 levels
2 bit- 4 levels
1 bit - 2 levels

How many different intensities (levels) can be used to create an image

Question 4

Image types

Answer 4

Binary, grayscale, color

Question 5

Point processing

Answer 5

Input is a single point, output is a single point

Question 6

Filtering

Answer 6

Input is a kernel of some size, output is a single pixel

Question 7

Darken using point processing

Answer 7

X-128

Question 8

Invert using point processing

Answer 8

255-x

Question 9

Lighten using point processing

Answer 9

X+128

Question 10

Lower contrast, raise contrast

Answer 10

X/2, x* 2

Question 11

Image histogram

Answer 11

Frequency of intensity value in an image

Question 12

Histogram equalization

Answer 12

Good for fixing overexposed or dark images so that more object captured within the image can be seen. Though in some cases it can decrease the image quality

Question 13

Histogram backprojection

Answer 13

Where in the image do the colors belong to the object being looked at

Question 14

Linear filtering

Answer 14

Form new image by replacing each pixel with a weighted sum of its neighbors using the same weights at each point

Question 15

Cross-Correlation

Answer 15

NEED MORE ONFO

Question 16

Filtering- identity

Answer 16

000
010
000

No change

Question 17

Filtering- shift

Answer 17

000
001
000

Shifted left by 1 pixel

Question 18

Filtering - blur

Answer 18

111
111 * 1/9
111

Box filter blur

Question 19

Filtering- sharpening

Answer 19

000 111
020 - 111 * 1/9
000 111

Sharpens images

Question 20

Convolution

Answer 20

Flip the filter (h) about the center and translate it (i,j) where filter need be applied. Then do dot product

Question 21

How to do deal with the filter “falling” off the edge

Answer 21

Clip filter (black) zero padding
Wrap around
Copy edge
Reflect across edge

Question 22

Types of noise

Answer 22

Salt and pepper noise (random b&w pix)
Impulse noise (Random white pixels)
Gaussian noise (variations in intensity from a Gaussian distribution)

Question 23

Box filter

Answer 23

1/9* 111
111
111

Question 24

Gaussian filter

Answer 24

1/16 * 121
242
121

Question 25

In Gaussian, theta is the

Answer 25

Standard deviation, and determines the effects of the smoothing, the higher it is the more spread out the smoothing curve is 3 is typically a good setting

Question 26

Median filtering

Answer 26

No liner filter Rank-order neighborhood intensities, Then take middle value Good for handling salt and pepper noise and other tiny scratches and weirdness unlike Gaussian that can smooth the image, but leads to greater image degradation