 Digital Image Acquisition

 Sampling and Quantization

 Image Representation
 Spatial, Temporal and Gray-Level Resolution
 Interpolation
 Pixel Neighbourhood
 Some Basic Relationships Between Pixels
 Neighborhood Adjacency and connectivity

 Connected Component Analysis

 Connected Component Labeling

 Distance Measures and Metrics

 Arithmetic and Logic Operations on Images

2
 EM Waves
 –A stream of mass less particles each travelling in a wave like pattern, moving at the speed of light and
contains a certain bundle of energy
 –The electromagnetic spectrum is split up in to bands according to the energy per photon

EM spectrum

3
 An image is typically generated by the combination of an “illumination” source and the
reflection, absorption, or radiation of energy by “scene”.

 An image is constructed as a function of illuminance and reflectance

5
 Computers cannot handle continuous images but only arrays of digital numbers.
 Thus it is required to represent images as two-dimensional arrays of points
 A point on the 2-Dgrid is called a pixel

6
projection image of object
through lens

7
projection onto
discrete sensor digital camera
array (CCD)
8
 AND QUANTIZATION

sampled image
sensors register
average color.

9
SAMPLING AND

discrete color output

continuous colors
mapped to a finite,
discrete set of colors.

10
continuous color input
 sampled &
real image sampled quantized
quantized

11
12
 Another Perspective
 The basic idea behind converting an analog signal to its digital signal is to convert both of its
axis (x,y) into a digital format.

 Since an image is continuous not just in its co-ordinates (x axis), but also in its amplitude (y
axis),
 The part that deals with the digitizing of co-ordinates is known as sampling.
 The part that deals with digitizing the amplitude is known as quantization.

13
 Another Perspective
 The concept of sampling is directly related to zooming.
 The more samples you take, the more pixels, you get.

 Quantization is opposite to sampling. It is done on y axis.

 On the x axis of the signal, are the co-ordinate values,
 On the y axis, we have amplitudes. So digitizing the amplitudes is known as Quantization.

14
Illustration of Sampling
Digital images consist of pixels.
On a square grid, each pixel
represents a square region of
the image.

The figure shows the same

image with
a) 3 × 4,
b) 2 × 16,
c) 48 × 64,
d) 92 × 256 pixels.

If the image contains sufficient

pixels, it appears to be
continuous.
16
Illustration of quantization

The same image is shown with

different quantization levels:
a 16,
b 8,
c 4,
d 2.

Too few quantization levels

produce false edges and make
features with low contrast partly
or totally disappear.

17
 An image is a 2D projection of a 3D world
 Image as a 2D function f(x,y): R2→R
 x and y are spatial coordinates on a rectangular grid (i.e. Cartesian coordinates)
 f at (x,y) is the intensity value

• A digital image is made

up of finite
elements/locations,
each having a finite
value

• Each element is called

a picture element
(pixel)

18
 A set of pixels (picture elements, pels)

 Pixel means
 –pixel coordinate
 –pixel value
 –or both

Both coordinates and

value are discrete

19
20
21
22
 Binary images:
 f(x,y) ∈ {0,1}

 Gray scale images

 f(x,y) ∈ C : C = {0,….,255}

 Color images
 Can be described mathematically as three gray scale images
 Red, green, blue channels
 fR(x,y) ∈ C, fG(x,y) ∈ C , fB(x,y) ∈ C
 C = {0,….,255}

 For digital images

 The minimum gray level value is usually 0
 The maximum depends on number of quantization levels used to digitize an image.
 The most common is 256 levels, so that the maximum gray level value is 255.

23
 Digital Image Acquisition
 Sampling and Quantization
 Image Representation
 Spatial, Temporal and Gray-Level Resolution
 Interpolation
 Pixel Neighbourhood
 Some Basic Relationships Between Pixels
 Neighborhood Adjacency and connectivity

 Connected Component Analysis

 Connected Component Labeling

 Distance Measures and Metrics

 Arithmetic and Logic Operations on Images

2
 An MxN image representation as a matrix
 Space required to store each intensity value
 Number of discrete intensity levels, L
 Number of bits used, k
L=2 k

 Intensity levels are equally spaced [0, L-1]

 Space required to store a digital image (in bits) M*N*k

24
 Color images have 3 values per
pixel; monochrome images have
1 value per pixel.

a grid of squares,
each of which
contains a single
color

each square is
called a pixel (for
picture element)

25
 Spatial resolution:
 The column (C) by row (R) dimensions of the image define the number of pixels used to cover the
visual space captured by the image.
 Relates to the sampling of the image information

 Temporal resolution
 For a continuous capture system such as video, this is the number of images captured in a given time
period.
 Measures in frames per second (fps); 30 fps normal video

 Bit resolution (Gray Level Resolution)

 This defines the number of possible intensity/colour values that a pixel may have.
 Relates to the quantization of the image information.
 Dynamic Range of Image
 Binary ? ; Grayscale ?? ; Colored ??

26
AND GRAY-LEVEL

Spatial Resolution

27
SPATIAL AND

256 128 16 8

64 32 4 2
28
 Zooming: Increasing the number of pixels in an image so that the image appears larger
 Zooming may be viewed as oversampling
 Shrinking may be viewed as undersampling.

 Interpolation is basic tool used extensively in tasks such as zooming, shrinking, rotation and
geometric corrections.
 Interpolation is a method of constructing new data points within the range of a discrete set of known
data points.
 Nearest neighbour interpolation
 Bilinear interpolation
 Bicubic interpolation

 Image shrinking: Subsampling

29
 Three methods

30
 Pixel replication

31
32
33
 Pixel Decimation

Decimation by a factor
of n:
take every nth pixel in
every nth row

34
35
 Nearest neighbour interpolation
 Simple but produces undesired artefacts

 Bilinear Interpolation
 Contribution from 4 neighbours

 Bicubic Interpolation
 Contribution from 16 neighbours

36
37
38
 Digital Image Acquisition
 Sampling and Quantization
 Image Representation
 Spatial, Temporal and Gray-Level Resolution
 Interpolation
 Pixel Neighbourhood
 Some Basic Relationships Between Pixels
 Neighborhood Adjacency and connectivity

 Connected Component Analysis

 Connected Component Labeling

 Distance Measures and Metrics

 Arithmetic and Logic Operations on Images

2
 Neighbors of pixel are the pixels that are adjacent pixels of an identified pixel

39
 Pixel Neighbourhood
N4(p) 4-neighbors: the set of horizontal and vertical neighbors
ND(p) diagonal neighbors: the set of 4 diagonal neighbors
N8(p) 8-neighbors: union of 4-neighbors and diagonal neighbors

O O O O O O
O X O X O X O
O O O O O O
N4(p) N8(p)
ND(p)

Pixel Neighbourhood coordinates explained

40
 Establishing boundaries of objects and components in an image
 Group the same region by assumption that the pixels being the same color or equal intensity
 Two pixels p & q are connected if
 They are adjacent in some sense
 If their gray levels satisfy a specified criterion of similarity

 Adjacency:
 Two pixels that are neighbors are adjacent if they have
 same grey-level or
 some other specified similarity criterion

 Pixels can be
 4-adjacent, diagonally adjacent, 8-adjacent, or m-adjacent.

41
 4-connectivity

 V: Set of gray levels used to define the criterion of similarity

Set of gray levels V = {1}

If gray level

42
 8-connectivity

 V: Set of gray levels used to define the criterion of similarity

Set of gray levels V = {1}

If gray level

43
46
 m-connectivity (Mixed Connectivity)

 V: Set of gray levels used to define the criterion of similarity

Set of gray levels V = {1}

If gray level

Note: Mixed connectivity can eliminate the multiple path connections that
often occurs in 8-connectivity 44
 Adjacency:
 Two pixels that are neighbors are adjacent if they have
 same grey-level or
 some other specified similarity criterion

 Pixels can be
 4-adjacent, diagonally adjacent, 8-adjacent, or m-adjacent.

 m-adjacency (mixed adjacency):

 Two pixels p and q of the same value (or specified similarity) are m-adjacent if either
i. q and p are 4-adjacent, or
ii. p and q are diagonally adjacent and do not have any common 4-adjacent neighbors.
iii. They cannot be both (i) and (ii).

45
 Connectivity in a subset S of an image
 Two pixels are connected if there is a path between them that lies completely within S.

 Connected component of S:
 The set of all pixels in S that are connected to a given pixel in S.

 Region of an image
 Boundary, border or contour of a region

 Edge
 A path of one or more pixels that separate two regions of significantly different
gray levels.

47
 Digital Image Acquisition
 Sampling and Quantization
 Image Representation
 Spatial, Temporal and Gray-Level Resolution
 Interpolation
 Pixel Neighbourhood
 Some Basic Relationships Between Pixels
 Neighborhood Adjacency and connectivity

 Connected Component Analysis

 Connected Component Labeling

 Distance Measures and Metrics

 Arithmetic and Logic Operations on Images

2
 In Matlab : bwconncomp(BW)
 Demo

48
4 Connectivity

49
50
Distance Measures
 Distance measures
 Distance function: A function of two points, p and q, in space that satisfies three criteria
( a ) D ( p, q ) ≥ 0
(b) D( p, q) = D(q, p), and
(c ) D ( p, z ) ≤ D ( p, q ) + D ( q, z )
p has coordinates (x,y)
 The Euclidean distance De(p, q) q has coordinates (s,t)
De ( p, q) = ( x − s ) 2 + ( y − t ) 2
2
2 1 2
 The city-block (Manhattan) distance D4(p, q) 2 1 0 1 2
D4 ( p, q ) =| x − s | + | y − t | 2 1 2
2 2 2 2 2 2
 The chessboard distance D8(p, q) 2 1 1 1 2
2 1 0 1 2
D8 ( p, q ) = max(| x − s |, | y − t |) 2 1 1 1 2
2 2 2 2 2 51
52
 Distance of each grey pixel from the closest white pixel.
 Which distance measures are used in the figure below?

53
 Arithmetic and logic operations are often applied a preprocessing steps in image analysis in
order to combine images in various way.

 Arithmetic Operations
 Addition,
 Subtraction,
 Division
 Multiplication

 Logic Operations
 AND
 OR,
 NOT

 These operation performed on two image ,

 except the NOT logic operation which require only one image, and are done on a pixel by pixel basis.

54
 Let x is the old gray value, y is the new gray value, c is a positive constant.
 Addition: y=x+c
 Subtraction: y = x - c
 Multiplication: y = cx
 Division: y = x/c
 Complement: y= 255 - x

 To ensure that the results are integers in the range [0, 255], the following operations should
be performed
• Rounding the result to obtain an integer
• Clipping the result by setting
• y = 255 if y > 255
• y=0 if y < 0

55
 MATLAB functions
 Addition: imadd(x,y)
 Add two images or add constant to image

 Subtraction: imsubstract(x,y)
 Subtract two images or subtract constant to image

 Multiplication: immultiply(x,y)
 Multiply two images or multiply image by constant

 Division: imdivide(x,y)
 Divide two images or divide image by constant

 Complement: imcomplement(x)

56
 Subtraction of two image is often used to detect motion.
 Consider the case where nothing has changed in a scene; the image resulting from subtraction of two
sequential image is filled with zeros - a black image.
 If something has moved in the scene, subtraction produce a nonzero result at the location of movement.

 Medical imaging
 Often uses this type of operation to allow the doctor to more readily see changes which are helpful in
the diagnosis.

 Law enforcement and military applications;

 for example, to find an individual in a crowd or to detect changes in a military installation.

57
 Subtraction of two images
 Change detection in Scene

58
 a) b)

c) d)

e) f)
Image Subtraction a) Original scene, b) same scene later, c) subtraction of scene a from scene b, d) the
subtracted image with a threshold of 50, e) the subtracted image with a threshold of 100, f) the subtracted image
with a threshold of 150.

Theoretically, only image elements that have moved should show up in the resultant image. Due to imperfect alignment between
the two images, other artifacts appear. Additionally, if an object that has moved is similar in brightness to the background it will
cause problems – in this example the brightness of the car is similar to the grass.
Added by 128 Subtracted by 128

60
 Noise is common
 G(x,y) = f(x,y) + n(x,y)

 Noise may be assumed as additive and uncorrelated

 Averaging may get rid of it.

 Digital subtraction angiography is commonly used to visualize vessels inside the body

61
 Used to adjust the brightness of an image.
 is done on a pixel by pixel basis and the options are to multiply or divide an image by a constant value,
or by another image.
 Multiplication of the pixel value by a value greater than one will brighten the image (or division by a value less than
1), and
 Division by a factor greater than one will darken the image (or multiplication by a value les than 1).

 Brightness adjustment by a constant is often used as a preprocessing step in image

enhancement

62
 a) original image, b) image divided by a value less than 1 to brighten, c)
image divided a value greater than 1 to darken

a) b) c)

63
 Image multiplication and division =>Shade Correction

64
Multiplied by 2 Divided by 2

65
Multiplication Addition

66
Division Subtraction

67
 The logic operations AND, 0R and NOT operate in a bit-wise fashion on pixel data.
 Example
 performing a logic AND on two images. Two corresponding pixel values are 11110 in one image and 8810
in the second image. The corresponding bit string are:

11110 = 011011112 88 = 010110002

011011112
AND 010110002
010010002

68
 AND and OR can be used as a simple method to extract a ROI from an image.
 For example,
 A white mask ANDed with an image will allow only the portion of the image coincident with the mask
to appear in the output image, with the background turned black;

 A black mask ORed with an image will allow only the part f the image corresponding to the black mask
to appear in the output image, but will turn the return of the image white.
 This process is called image masking

69
a) b) c)
a) Original image,
b) image mask for AND
operation,
c) Resulting image from
(a) AND (b),
d) image mask for OR
operation, created by
performing a NOT on
mask (b),
e) Resulting image from
(a) OR (d).
d) e) 70
 Mask application

71
 Create the negative image
 MATLAB:
 function:
 x = imread(‘filename.ext’);
 y = imcomplement(x);

72
 Digital Image Processing (3rd Ed) By Gonzalez
 Chapter 2 ( some sections )

 Digital Image Processing using Matlab

 Chapter 2
 For Introduction of Matlab and basic image processing operations,

 Fundamental of Digital Image Processing A practical approach with examples in Matlab

(www.fundip.com )
 Chapter 3

73
Various contents in this presentation have been taken from different books, lecture notes, and
the web. These solely belong to their owners, and are here used only for clarifying various
educational concepts. Any copyright infringement is not intended.

74