Beruflich Dokumente
Kultur Dokumente
- Image
- Digital Image
Image:
- An image is a two-dimensional picture,
that has a similar appearance to some subject;
- A physical object or a person
- Two-dimensional images:
- A photograph
- A screen display
- A drawing/ painting
- Image can also be a three-dimensional, such as:
- A hologram
- Stereoscopic 3-D image
- 3D movie
Capturing Images :
- Optical devices:
- Cameras
- Lenses and mirrors
-Telescopes and Microscopes etc
- Natural objects and phenomena:
- The human eye or
- Water surfaces, Rain droplets
Digital Images
- Generally images are analog (Taken by a roll film cameras/ paintings)
- For computer processing these images have to be digitazed
- Digitized images:
- Images converted to digital image form by
- Digitizers/ Scanners
- Images acquired by digital cameras
- Produces digital images, directly
- Digital images are represented by pixels (Picture elements)
- A digital image is defined as
a two dimensional function f (x, y)
where x and y are spatial ( plane ) coordinates
f is amplitude of light intensity/ or grey level
at any pair of coordinates (x, y)
(Average value of the grey level or intensity of light
of the pixel area surrounding the point (x, y))
Human Vision
- Most advanced of our senses
- Plays most important role in Human perception
- Limited to only visible band of spectrum
( Electro magnetic spectrum)
Electromagnetic spectrum
Frequency (Hz)
Wavelength (Meters)
Visible Spectrum
CD
DVD
Blue ray disc
Electromagnetic waves
- X-Rays
- Ultra violet rays
- Visible range
- Infra red
- Microwaves
- Radio waves
- Sound waves
Ultra sound
0.63 to0.69
- Vegetation discrimination
- Microwaves
Radar
- Works in ambient light and in any weather
- Penetrates clouds
- See-through ice/ dry sand
- Radio waves
- Radio Astronomy
- MRI ( Magnetic Resonance Imaging )
- Sound waves
- Acoustics
100 of Hertz
- Geographical Exploration/ Oil exploration
Industry
- Ultrasonic
Million of Hertz
Pulse Echo
- Seismic images
- Medical applications
- Electron beams
Used in electron microscopy
10,000 x amplification
- Computer generated Images
- Synthetic images
Used for 3-D modeling and visualization
Virtual Reality
- Fractal Images
Fractals is iterative reproduction of basic pattern
according to some mathematical rule
X-ray images
Angiogram
Circuit board
CT scan: Head
Computed
tomography
Thumb
print
Paper currency
Baby image
Ultra sonic
Fractal images
(compression)
b. Transmission
(compression)
c. Representation and
d. Automatic machine perception
2. Image Analysis:
Understanding of image characteristics
Image in Measurement out
3. Computer vision:
To use computers to emulate human vision
- Recognizing objects in images
Being able to draw inferences, take actions
Image in High level description out
Types of sensors:
- Single sensor:- One sensor for all pixels
A Mechanical system moves over all the parts of
a picture in front of the single sensor
Array of Sensors:
Array of Sensors
located at image plane
Object
Continuous image
Sampling:
Quantization:
Digitization of an Image
y
Pixel 0, 0
N = 16 columns
x
Value of intensity of light
or grey level
0 255 for 8-bit
Pixel coordinates
x = 3, y = 10
M = 16
rows
Digitization of an image
1. Place a grid on the image
2. Divide the image into picture elements (pixels)
3. Measure the average value of grey level at
Each grid element/ picture element ( pixel) and
Quantize the grey level value
(by mapping into the specified grey level scale)
Digital image is a function f (x, y)
where x = 0 to M-1 and y = 0 to N-1
f (x, y) is the value of pixel at (x, y), in integers (quantized)
(0 to 255 for 8-bit values)
Digital image
Y coordinate, 0 to N-1
X
coordinate
0 to M-1
f(0, N-1)
f(1, N-1)
f(M-1, N-1)
Parameter
Symbol
Typical values
Rows
Columns
Gray Levels
(Bits
L
B
Generally:
M = N = 2B
where {B = 2, 4, 8, 10,12, 16}
L = 2 8 (8 bit, or 1 byte)
One colour
image is
split into
three
grey scale
images
by using filters
2. Choroid
- Lies directly below the sclera
- It is a network of blood vessels
- Choroid serves as a source of nutrition to the eye
- It is heavily pigmented
Which helps reduce amount of
extraneous light entering the eye and also
reduces the back scatter within the optic glove
- At the anterior (front) extreme, the Choroid is divided into:
1. Ciliary body and
2. Iris
- Iris
- Iris contracts or expands to control the amount of light
that enters the eye
- The central opening of iris varies from 2 to 8 mm
- The front of the iris contains the visible pigment of eye
- - black/ brown pigment
- Lens
- Layers of fibrous cells
attached to the Ciliary body
(Cataracts - clouding of lens)
- Focal length of the lens 14 mm to 17 mm)
3. Retina
- Innermost membrane of eye
- lies inside of the walls posterior portion
- When the eye is focused,
light from the object outside the eye
is imaged on the retina
- A distribution of discrete light receptors over
the surface of retina sense the pattern of light
- There are two classes of receptors:
1. Cones and
2. Rods
:- Cones:
- About 6-7 million
- Primarily in central portion of retina
- Cones are sensitive to color
- Eye rotates till the image is focused on the cones area
- Eye resolves fine details of an image with the Cones,
- Cone vision is called: Photopic or Bright-light vision
- Rods:
- 70 -150 million, distributed over the backside of retina
- Rods gives general overall picture of fields of view
- Sensitive to overall intensity of light, not colours
- These are sensitive to low levels of illumination
- Rod vision is called:
3. Retina
Lens Focal
Length 14 18 mm
Blind spot
1. Cornea (Outer
transparent cover)
Fovea
Visual axis
Iris diaphragm
2 - 8 mm
Central opening
Vitreous humor
Diameter = 20 mm
Ciliary muscle
Ciliary fibers
h / 17 = 15/100
h = 17 x 15/100 = 2.55 mm
32 x 32 pixel size
128 level
image
64 level
image
32 level
16 level
image
8 level
image
4 level
image
2 level
image
x -1
y -1
y +1
x +1
x-1, y-1
x+1, y+1
Adjacency
Let V be the set of intensity values, used to define adjacency
2. 8-adjacency:
Two pixels p and q are in 8-adjacency
with value V (subset of 0-256 values)
If q is in set N 8 (p)
p
3. Mixed adjacency
( m-adjacency )
Modification of 8 adjacency
Two pixels p and q are in m-adjacency
with value V (subset of 0-256 values) / or 1
1. If q is in set N4(p)
p
N4(p)
or
2. if q is in set Nd (p)
and
Nd(p)
q
p
q
p
Pixels
in Intersection
8 - Adjacency
two paths
M - path
Intersection
pixel 1
q1
Intersection
pixel 0
q2
M-adjacency:
-A mixture of two, N4 and Nd adjacencies to
eliminate ambiguity of 8 -Adjacency
- If q is at diagonal point then none of the two pixels
across diagonal positions should be 1
- In diagonal path there should be no straight path
Since q1 is at diagonal and intersection of P4 and D4 is 1 so the diagonal path is
not possible
Since q2 is at diagonal and intersection of P4and D4 is 0 so the diagonal
path is possible
i-1
1in
Connected path in S
S
q
Connected pixels
Connected componant
- For any pixel p in S,
The set of pixels, that are connected to the pixel p in S,
is called a connected component of S
Connected set
- If it (S) has only one connected component
then set S is called connected set
Region
- Let R , a subset of pixels in an image,
R is called a region of the image
if it is is a connected set
Adjacent regions
- Two regions are said to be adjacent
if their union forms a connected set
- Adjacency can be 4 or 8
Disjoint regions
- Two regions that are not adjacent. (Having no 1s connected)
111
101
010
001
111
111
R i region
R j region
- Ri and Rj are adjacent, if 8-adjacency is considered
- Ri and Rj are disjoint, if 4-adjacency is considered
since 4-path does not exist, between Ri and Rj
-Edge
Difference between the edge and the boundary:
- Boundary of a region forms a closed path
it is a global concept
Distance measure
z (v, w)
Two distances:
1. Euclidean distance
2. Block level distance
q (s, t)
p (x, y)
D is distance function or metric if:
a.
D (p, q) is positive 0
b.
D( p, q) = D ( q, p)
d.
D ( p, z) D( p, q) + D (q, z)
( D( p, q) = 0 if
D e = [ (x s) 2 + (y t) 2 ]1/2
For this distance measure,
the pixels having a distance some value r from (x, y)
are the points contained in a disk of radius r
p q)
r
x, y
q (s, t)
p1
p0
p3 p 4
p2
p3 p 4
p1 p 2
p0
(1)
1. If p3 and p1 are 0s
2. if only p1 = 0
3. if only p3 = 0
4. if p1 and p4 are 1
p1
p0
p3 p 4
p2
(2)
p1
p0
p3 p 4
p2
(3)
p3 p4
p1 p2
p0
(4)
(p0, p2, p4)
(p0, p2, p3,p4)
(p0, p1, p2, p4)
(p0, p1, p2, p3, p4)
Determine
1. Euclidian distance
2 . City block distance
3. Chess board distance between p and q in the following subimage)
1 2 3 4 5 6 7 8
1
2
3
4
5
6
7
8
9
10 p
p = (10, 1)
q = ( 2, 8)
a12
a21
a22
b11 b12
b21 b22
a11 x b11
a12 x b12
a21 x b21
a22 x b22
Matrix product:
a11
a12
b11 b12
a21
a22
b21 b22
i=k
Average of g (x, y) = 1/ k
g i (x, y)
i =1
Where k is the number of images
- Astronomy:
- Low light levels images are totally noisy
- By adding multiple images noise can be reduced
10
Noisy image
20
50
100
Other images are result of averaging over 5, 10, 20, 50 and 100 images
- Average of grey levels for each pixel in the images
Scaling
-Images after arithmetic operations may go out of range of levels
(0 -255)
to < 0 or > 255
- First the minimum value of intensity in the image is brought to zero:
f m = f min (f)
(Intensity of every pixel Minimum intensity of pixels)
- Then scale image pixel intensities by
f s = K x ( f / max (f m )
where K is the maximum level of grey scale (255)
- Scaled value of any pixel = grey value of the pixel multiplied by
255/ maximum grey value of the pixels
in the image
- So that the intensity range of the image becomes 0 - 255
Image A
30 50
50 70
20 80
100 230
80 200
90 180
Image B
20 30
40 60
10 50
120 240
90 200
80 170
A+ B
50 80 220 470
90 130 170 400
30 100 160 350
Minimum intensity = 30
Scaled image A +B
12 29
35 58
00 41
110 255
82 215
75 186
20 50
60 100
00 70
190 440
140 370
130 320
Two images are subtracted then scaled to the intensity levels up to 255
- Image is obtained by taking difference of the two images
- Then it is scaled to the full range of grey values (0 to 255)
Original image
Image obtained
by setting the least
significant bit to 0
Image obtained
by scaling the image
to 0 to 255
Difference between
Mask and
Live image
Live image
Enhanced difference
image
Shading pattern
Processed image
- Product of original image
and
reciprocal of shading pattern
Logical operations:
Binary images
1 valued pixels as foreground
0 valued pixels as background
Union of two binary images
OR
NOT
Original mage
Original mage
OR image mask
Result of OR
OR 0 0 = 0
operation 1 1, 1 0, 0 1 = 1
End