Digital Image Processing

LECTURE: 03,04
(CH 02)
 Spatial and Intensity Resolution


Spatial and Intensity Resolution (Cont.)

 Example: A digital camera with a 20-megapixel CCD

imaging chip can be expected to have a higher
capability to resolve detail than an 8-megapixel
camera,
 Assuming that both cameras are equipped with comparable
lenses and
 The comparison images are taken at the same distance.
 Intensity resolution similarly refers to the smallest
discernible change in intensity level.
Spatial and Intensity Resolution (Cont.)
 Image Formation

Courtesy: DIP 3/Edition

Umer Javed, CS: 407: Digital Image Processing Chapter:2, Slide # 5
 Image Formation (Cont.)

Image Formation (Cont.)
 Image Sensing
 Mostly images are generated by combination of the

“illumination source” and the

“reflection/absorption/transmission” of the energy by the

element of scene.
 Reflected or transmitted energy is used/directly by sensor plane/film to generate image

 Some cases Reflected or transmitted energy is focused on a photoconverter to convert it into

visible light.

 This visible light is detected by another sensing device to create digital image (Normally

digital imaging)
 Image Sensing (Cont.)
 Principal Sensor Arrangement:
 Single image sensor (Fig 1).

 Linear sensor (Fig. 2)

 Array sensor (Fig. 3)

 Image Sensing (Cont.)
 Principal Sensor Arrangement:
Fig. 1

Fig. 3
 Image Sensing (Cont.)

 Imaging using single sensor

 Photodiode follows this kind of behavior.

 Colored sensors give maximum voltage for similar

colors and minimum for others.

Chapter:2, Slide # 11
 Image Sensing (Cont.)

Fingerprint sweep sensor

Computerized Axial Tomography
 Image Sensing
 Sensor Arrays:

 Charged Coupled Devices (CCD)

 Widely used in digital cameras

 Complimentary Metal Oxide Semiconductor (CMOS)

 CMOS doesn’t require charged couples

 Application Areas:

 Air Borne Imaging

 X-Rays Scan

 Computerized Axial Tomography (CAT)

 Positron Emission tomography (PET)

 Magnetic Resonance Imaging (MRI)

 Image Sensing (Cont.)

w Used for convert a continuous

image into a digital image

w Contains an array of light sensors

w Converts photon into electric charges

accumulated in each sensor unit

CCD KAF-3200E from Kodak.

(2184 x 1472 pixels,
Pixel size 6.8 microns2)
 Pixel relationships
(0,0) y
 Neighborhood

 Adjacency (x-1,y-1) (x-1,y) (x-1,y+1)

 Connectivity (x,y-1) (x,y) (x,y+1)

x

 Paths
(x+1,y-1) (x+1,y) (x+1,y+1)

 Regions and boundaries

Chapter:2, Slide # 15
 Spatial Relationship
 Nearest Neighborhood

 4-Connected pixels are neighbors to every pixel that touches one of

their edges.

 These pixels are connected horizontally and vertically.

 N4 = (x + 1, y),(x − 1, y),(x, y + 1),(x, y −1)

 Diagonal Neighbors

ND = (x + 1, y + 1),(x + 1, y − 1),(x − 1, y + 1),(x − 1, y − 1)

 Eight Neighbors

N8 = N 4 ND

Chapter:2, Slide # 16
 Spatial Relationship

Pixel neighborhood: (a). Nearest/N4, (b). Diagonal/ND and (c). N8 neighborhood

Chapter:2, Slide # 17
 Pixel adjacency
 Let v={}, set of intensity values

 Binary image, v={1 or 0}.

 Determine adjacency with “1” intensity pixels.

 Similarly it can be done for gray scale images.

 Types of adjacency

 4-adjacency

 8-adjacency

 M-adjacency

Chapter:2, Slide # 18
 Pixel adjacency
 4-adjacecncy:

 Two pixels ‘p’ and ‘q’ are 4-adjacent IF

 ‘p’ and ‘q’  set ‘v’ AND ‘q’ is in N4(p).

 8-adjacecncy:

 Two pixels ‘p’ and ‘q’ are 8-adjacent IF

‘p’ and ‘q’  set ‘v’ AND ‘q’ is in N8(p).

 M-adjacency:

 Two pixels ‘p’ and ‘q’ are M-adjacent IF ‘p’ and ‘q’  set ‘v’
AND ‘q’ is in N4(p).

 ‘q’ is in ND(p) AND N4(p) N4(q) has no pixel  set ‘v’

 Path

q
p

We can define type of path: 4-path, 8-path or m-path

depending on type of adjacency.
 Pixel adjacency
 Given, v={1 , 2} the adjacent pixels and path are:

Chapter:2, Slide # 21
 Connectivity


0  i  n, (x i , yi )  S i

Chapter:2, Slide # 22
 Connectivity
 Let S represent a subset of pixels in an image

 For every pixel p in S, the set of pixels in S that are connected to p is

called a connected component of S.

 If S has only one connected component, then S is called Connected

Set.

 We call R a region of the image if R is a connected set.

 Two regions, Ri and Rj are said to be adjacent if their union forms a

connected set.

 Regions that are not to be adjacent are said to be disjoint.

Chapter:2, Slide # 23
 Border, Background and Foreground
 Boundary (or border)

 The boundary of the region R is the set of pixels in the region that

have one or more neighbors that are not in R.

 If R happens to be an entire image, then its boundary is defined as the

set of pixels in the first and last rows and columns of the image.

 Foreground and background

 An image contains K disjoint regions, Rk, k = 1, 2, …, K.

 Let Ru denote the union of all the K regions, and let (Ru)c denote its

complement.

 All the points in Ru is called foreground,

 All the points in (Ru)c is called background.

Chapter:2, Slide # 24
 Mathematical operations

 Following mathematical operations can be

performed on digital
 Linear or Non-linear operations

 Arithmetic operations

 Spatial operations

 Distance measurements

 Geometric transformations

Chapter:2, Slide # 25
 Linear / Non linear operations

 Linear operator need to satisfy two condition

 Additivity

 Homogeneity

Chapter:2, Slide # 26
 Arithmetic Operations

 Two images A and B can be:

 Added

 Subtracted

 Multiplied

 Divided

Chapter:2, Slide # 27
 Arithmetic Operations
 Averaging for noise reduction.
 Arithmetic Operations
Addition for dynamic contrast enhancement
[Debevec, Malik, 1997]

Chapter:2, Slide # 29
 Arithmetic Operations

 Subtraction for highlighting differences

Chapter:2, Slide # 30
 Arithmetic Operations

 Image subtraction for Biomedical imaging.

In Angiography, subtraction of new frame from

previous highlights the difference. After this, contrast
enhancement can also be used.

Chapter:2, Slide # 31
 Arithmetic Operations
 Image multiplication for segmentation.

Chapter:2, Slide # 32
 Arithmetic Operations
 Image multiplication for segmentation.
 Arithmetic Operations
 Set operations.
 Arithmetic Operations

Courtesy: DIP 3/Edition

Umer Javed, CS: 407: Digital Image Processing Chapter:2, Slide # 35

 Distance Calculations
 Given pixels p, q and z with coordinates (x, y), (s, t),
(u, v), respectively.
 The following distances can be found:
 Euclidean distance:
1
2 2 2
DE (p, q) = [(x - s) + (y - t) ]

 City Block distance:

D4 (p,q) =| x - s | + | y - t |

 Chess Board distance:

D8 (p, q) = max(| x - s |,| y - t |)
 Image Interpolation

Original Image Resample Image Zoomed Image

 Enhancing Resolution

Low Resolution

High Resolution
Inserting Rows and Columns
 Linear Interpolation
Filling of new locations by using linear interpolation method

Note: a=0.5, will result in average of two

f(n)=[0,120,180,120,0]
f(x)=[0,60,120,150,180,150,120,60,0], x=n/2
f(x)=[0,20,40,60,80,100,120,130,140,150,160,170,180,…], x=n/6
Courtesy: image_interpolation.ppt, google.com

Umer Javed, CS: 407: Digital Image Processing Introduction 39

Chapter:2, Slide # 39
 Higher Order Interpolation

Filling of new locations by using non linear interpolation method

Courtesy: image_interpolation.ppt, google.com

Introduction 40
Chapter:2, Slide # 40
 Image Interpolation
 Nearest neighborhood.
 Bilinear Interpolation. (Set N=1)
 Bicubic Interpolation. (Set N=3)

Chapter:2, Slide # 41
Bilinear Interpolation

Courtesy: image_interpolation.ppt, google.com

Chapter:2, Slide # 42
Technique comparisons
 Bilinear Interpolation

100 x 100

400 x 400
 Bicubic Interpolation

100 x 100

400 x 400
Courtesy: image_interpolation.ppt, google.com

Chapter:2, Slide # 45
 Affine Transformation
 It is an important class of linear 2-D geometric
transformations.
 It maps the variables of an input image into new
variables (e.g. in an output image) by applying a
linear combination of following operations.
 Translation

 Scaling

 Rotation

 Shear
Chapter:2, Slide # 46
Translation

Courtesy: S. Savarese

Chapter:2, Slide # 47
 Translation

 Alternative approaches,

Courtesy: S. Savarese

Chapter:2, Slide # 48
Scaling

Courtesy: S. Savarese

Chapter:2, Slide # 49
Translation and Scaling

Courtesy: S. Savarese

Chapter:2, Slide # 51
 Translation and Scaling

 Alternative approach,

Courtesy: S. Savarese

Chapter:2, Slide # 52
Rotation

Courtesy: S. Savarese

Chapter:2, Slide # 53
 Multiple operations
 Translation, scaling and rotation.

Courtesy: S. Savarese

Chapter:2, Slide # 54
Summary