CV #2 Digital Image, Fundamentals

Computer Vision
2023-24 First Semester, M.Tech (AIML)
Session #2:
Digital Image Fundamentals
Instructors, CV course
S. P. Vimal | CSIS Department | WILP | BITS – Pilani (vimalsp@wilp.bits-pilani.ac.in)

Acknowledgement: Slide Materials adopted from - Intro to Computer Vision (Cornell Tech); Noah Snavely and prepared by the lead instructor
Prof S P Vimal
Topics
● Image Representation
● Image Digitization ( Sampling & Quantization)
● Digital Image Properties
Acknowledgement: Slide Materials adopted from - Intro to Computer Vision (Cornell Tech); Noah Snavely
2
3
4
5
6
Image representation for analysis
7
Representations
8
3D vision tasks and algorithmic components
9
Ref: https://medium.com/@nrmin.mrdova.01/pinhole-camera-842ce79329b7
10
Pinhole Camera inverting image
11
Simple Image Formation Model
12
Image Formation in the Eye
13
Digital Image Acquisition
14
Quiz
Why do we use lenses when we have pin-hole camera?
Cameras with lenses are easier to manufacturer when compared to pin-hole cameras
Lenses gather more light than pinhole cameras, reducing exposure time
Lenses can be stacked to produce a clear image of the scene
Lenses keep parallel lines from the scene, parallel in the image
15
Simple Image Formation Model
● We denote images by two-dimensional functions of the form

f(x, y).
○ Value of f at spatial coordinates (x, y) is a scalar quantity;
○ Physical meaning of f(x,y) is determined by the source of the image, and whose
values are proportional to energy radiated by a physical source (e.g.,
electromagnetic waves);
● Function f(x, y) is characterized by two components:

○ Amount of source illumination incident on the scene being viewed i(x,y), and
○ Amount of illumination reflected by the objects in the scene r(x,y).
16
0 is total absorption and 1 is total reflectance
Simple Image Formation Model - Grayscale Image
● Let the intensity (gray level) of a monochrome image at any

coordinates (x, y) be denoted by
● For a gray scale, the range is scaled between [0, L-1] ;

○ All intermediate values from 0 to L-1 are shades of gray varying from black
to white.
17
Sampling and Quantization
Continuous image A scan line showing intensity variations

along line AB in the continuous image
Sampling and quantization. Digital scan line
18
Digitizing the coordinate values is called sampling ; Digitizing the amplitude values is called quantization.
Sampling and Quantization
Continuous image projected Result of image sampling and

onto a sensor array quantization
19
Matrix Representation of an Image
20
Spatial & Intensity Resolution
930 dpi 300 dpi

● Spatial resolution is a measure of
the smallest discernible detail in
an image;
● Intensity resolution similarly refers

to the smallest discernible change
in intensity level
150 dpi 72 dpi
21
256-level 128-level
● Spatial resolution is a measure
of the smallest discernible detail
in an image;
● Intensity resolution similarly

refers to the smallest discernible
change in intensity level
64-level 32-level
All images have spatial resolution 2022 × 1800 22

16-level 8-level
● Spatial resolution is
a measure of the
smallest discernible
detail in an image;
● Intensity resolution
similarly refers to the
smallest discernible
change in intensity 4-level 2-level
level
All images have spatial resolution 2022 × 1800 23

Distance measurement
24
Distance Measurement
25
What is an image?
A grid (matrix) of intensity values
26
What is an image?
Can think of a (grayscale) image as a function f from R2 to R:

f (x,y) gives the intensity at position (x,y)
27
Image Transformations
As with any function, we can apply operators to an image
28
Image Histogram
Un-normalized histogram :
Note: nk is the number of pixels in f with intensity rk
Normalized histogram :
Note: M and N are the number of image rows and columns

The sum of p(rk) for all values of k is always 1
Mostly, we work with normalized histograms
29
Elementary Intensity Transformations
A digital mammogram and its Negative image

30
Log Transformations
Fourier spectrum displayed as a

grayscale image and the result of
applying the log transformation in
with c = 1.
Both images are scaled to the

range [0, 255].
31
Log Transformations
Fourier spectrum displayed as a

grayscale image and the result of
applying the log transformation in
with c = 1.
Both images are scaled to the

range [0, 255].
32
Power-Law Transformations
● c and ˠ are positive constants
33
Power-Law Transformations- Applications
Gamma correction
● Cathode ray tube (CRT) devices have an
intensity-to-voltage response that is a
power function, with exponents varying
from approximately 1.8 to 2.5.
34
Gamma correction
● Gamma correction consists of using the

transformation
35
Gamma correction
● Gamma correction consists of using the

transformation
36
Medical Imaging
(a) Magnetic resonance image (MRI) of a
fractured human spine (the region of the fracture
is enclosed by the circle).
(b)–(d) Results of applying the transformation

with c = 1 and g = 0.6, 0.4, and 0.3, respectively.
(Original image courtesy of Dr. David R. Pickens,
Department of Radiology and Radiological
Sciences, Vanderbilt University Medical Center.)
37
Aerial Imaging
(a) Aerial image.

(b)–(d) Results of applying the transformation in with 𝛄 =
3.0, 4.0, and 5.0, respectively. (c = 1 in all cases.)
(Original image, courtesy of NASA.)
38
Contrast Stretching; Thresholding
(a) Piecewise linear transformation function.

(b) A low-contrast electron microscope image of pollen,
magniﬁed 700 times.
(c) Result of contrast stretching.
(d) Result of thresholding.
39
Contrast Stretching
40
Intensity Level Slicing
Highlights range [ A, B] and reduces This function highlights range [ A, B] and

all other intensities to a lower level. leaves other intensities unchanged.
41
• Readings:
Digital Image Processing, Rafael C. Gonzalez & Richard E woods, Third Ed, Chapter 2
References for hands-on work:

https://www.geeksforgeeks.org/log-transformation-of-an-image-using-python-and-openc
v/
https://www.kaggle.com/code/gauravduttakiit/image-processing-notebook
42
Thank you
43

CV #2 Digital Image, Fundamentals

Uploaded by

Copyright:

Available Formats

CV #2 Digital Image, Fundamentals

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

CV #2 Digital Image, Fundamentals

Uploaded by

Copyright:

Available Formats

Computer Vision

2023-24 First Semester, M.Tech (AIML)

S. P. Vimal | CSIS Department | WILP | BITS – Pilani (vimalsp@wilp.bits-pilani.ac.in)

Why do we use lenses when we have pin-hole camera?

● We denote images by two-dimensional functions of the form

● Function f(x, y) is characterized by two components:

● Let the intensity (gray level) of a monochrome image at any

● For a gray scale, the range is scaled between [0, L-1] ;

Continuous image A scan line showing intensity variations

Sampling and quantization. Digital scan line

Continuous image projected Result of image sampling and

930 dpi 300 dpi

● Intensity resolution similarly refers

● Intensity resolution similarly

All images have spatial resolution 2022 × 1800 22

All images have spatial resolution 2022 × 1800 23

A grid (matrix) of intensity values

Can think of a (grayscale) image as a function f from R2 to R:

As with any function, we can apply operators to an image

Note: nk is the number of pixels in f with intensity rk

Note: M and N are the number of image rows and columns

A digital mammogram and its Negative image

Fourier spectrum displayed as a

Both images are scaled to the

Fourier spectrum displayed as a

Both images are scaled to the

● c and ˠ are positive constants

● Gamma correction consists of using the

● Gamma correction consists of using the

(b)–(d) Results of applying the transformation

(a) Aerial image.

(Original image, courtesy of NASA.)

(a) Piecewise linear transformation function.

Highlights range [ A, B] and reduces This function highlights range [ A, B] and

References for hands-on work:

You might also like