IMAGE TRANSFORMATION IN DIGITAL IMAGE

PROCESSING
PART-B
GATE GEOMATICS ENGINEERING

IMAGE
TRANSFORMATION

IMAGE
PCA TASSELED CAP COLOUR SPACE FOURIER
AIRTHMATIC IMAGE FUSION
TRANSFORMATION TRANSFORMATION TRANSFORMATION TRANSFORMATION
OPERATIONS
 Digital Image
Processing

Image Image Image

Pre-Processing
Enhancement Transformation Classification
 IMAGE
TRANSFORMATION

IMAGE AIRTHMATIC PCA TASSELED CAP COLOUR SPACE FOURIER

IMAGE FUSION
OPERATIONS TRANSFORMATION TRANSFORMATION TRANSFORMATION TRANSFORMATION
 IMAGE ADDITION

IMAGE AIRTMATIC
IMAGE

OPERATION
SUBTRACTION

IMAGE
MULTIPLICATION

INDICES/RATIONING
 Image Transformation
Image transformations generate ‘new’ images from two or more sources which
highlight particular features or properties of interest, better than the original input
images.

Common transformations:
Image arithmetic operations
Principal component transformation (PCT) Tasselled cap
transformation (TCT) Colour space transformation (CST)
Fourier transformation
Image fusion
Image Arithmetic Operations (Image Addition)
Addition is getting new DN value of a pixel in the output image by
averaging the DN values of corresponding pixels of input images.

(16 + 56) / 2

16 18 26 58 28 56 19 23 56 23 36
16 15 13 23 46 19 14 16 24 43
78 23 13 93 36 75 25 16 98 34
72 33 64 25 26 73 32 61 20 28
18 88 36 72 88 13 85 37 72 89
Input 1 Input 2 Output

Example: To derive the average temperature

 Image
Subtraction
It is the process of subtracting the DN value of one image from
corresponding pixel of other image.

2 - 7

2 5 4 6 7 5 3 1 -5 0 1 5
3 5 8 9 1 9 3 0 2 -4 5 9
6 7 9 5 6 9 9 3 0 -2 0 2
8 9 6 8 8 6 2 7 0 3 4 1
Input 1 Input 2 Output
 Change
Detection
1987 1997
 Image
Multiplication
In this process pixel-by-pixel multiplication of two images is performed.

2 × 2

2 5 4 6 2 5 4 6 4 25 16 36
3 5 8 9 3 5 8 9 9 25 64 81
6 7 9 5 6 7 9 5 36 49 81 25
8 9 6 8 8 9 6 8 64 81 36 64
Input 1 Input 2 Output
Ratios
• The ratio of two bands removes much of the effect of
illumination in the analysis of spectral differences. This is
possible because the composition differences between two
geologic units (here limestone, which is “gray” in RGB321,
and a “red bed” which is red in RGB321) determine the ratio
of any two bands, whereas the illumination determines the
magnitude of the DN received by the satellite..
 • Used to create output image by
mathematically combining the DN values
of different bands.
• Spectral indices are combinations of
spectral reflectance from two or more
wavelengths that indicate the relative
abundance of features of interest.
Indices Vegetation indices are the most popular
type, but other indices are available for
burned areas, man-made (built-up)
features, water, and geologic features.
• For example
• Vegetation index : DNNIR / DNR
 Concept of Indices
VEGETATION
TM Band 4: HIGH Value
TM Bands 1 2 3 4
TM Band 3: LOW Value
Vegetation
60
TM Band 4 ÷ TM Band 3
% Reflectance
Dry Soil = HIGH VALUE

DRY SOIL

TM Band 4: MODERATE Value

0 TM Band 3: MODERATE Value

0.4 0.7
Wavelengths in microns
TM Band 4 ÷ TM Band 3
= LOW VALUE
Concept of Indices
 Concept of Indices
Common indices:

Vegetation index DNNIR / DNR

Normalized differential vegetation index (DNNIR – DNR) / (DNNIR + DNR)

Soil adjusted vegetation index [(DNNIR – DNR) / (DNNIR + DNR + L)] × (1+L)

Transformed vegetation index [{(DNNIR – DNR) / (DNNIR + DNR)} + 0.5]½

Perpendicular vegetation index [(DNR – DNR ) + (DNNIR – DNNIR )]

soil veg soil veg ½

Normalized difference snow index (DNG – DNSWIR) / (DNG – DNSWIR)

 Normalized Differential
Vegetative Index

Stressedvegetation
Moderate+vevalues

Healthyvegetation
High+vevalues

Range of values: -1 to +1
 IMAGE
TRANSFORMATION

IMAGE AIRTHMATIC PCA TASSELED CAP COLOUR SPACE FOURIER

IMAGE FUSION
OPERATIONS TRANSFORMATION TRANSFORMATION TRANSFORMATION TRANSFORMATION
 The objective of PCT is to reduce the dimensionality
of data and compress as much of the information in
the original bands into fewer bands.

Principal
The ‘new’ bands that result from PCT are called
Component principal components.

Transformation
PCT attempts to statistically maximize the amount of
information (or variance) from the original data into
the least number of new components.
 Image Space and
Scatterplot

5 7 1 2 3 4 P1 P2 P3 P8

Band 2
P3 P5 P6 P7
4 2 3 1 4 4 P4 P5 P6 P9 P2
P1
5 2 4 4 5 3 P7 P8 P9
P4

Band 1 Band 2 Pixels

Band 1
Two-band Scatterplot
Three-band Scatterplot

Band2

Band1
1. Center the data
2. Calculate the covariance matrix
3. Calculate eigenvalues of the covariance matrix
4. Calculate eigenvectors of the covariance matrix
5. Order the eigenvectors
6. Calculate the principal components
Principal Component
Transformation
 Principal Component
Transformation
B-1 B-2 B-3

B-4 B-5
 Principal Component
Transformation
PC-1 PC-2 PC-3

PC-4 PC-5
 IMAGE
TRANSFORMATION

IMAGE AIRTHMATIC PCA TASSELED CAP COLOUR SPACE FOURIER

IMAGE FUSION
OPERATIONS TRANSFORMATION TRANSFORMATION TRANSFORMATION TRANSFORMATION
 Tasselled Cap Transformation

Used for the analysis of crop growth

Based on linear transformation

(kauth and Thomas 1976)
 Tasselled Cap Transformation

Landsat-MSS
 TasselledCapTransformation
• Landsat TM

• Redness = 0.3037 TM1 + 0.2793 TM2 + 0.4743 TM3 + 0.5586 TM4 + 0.5082
TM5 + 0.1863 TM7

• Greenness = –0.2848 TM1 – 0.2435 TM2 – 0.5436 TM3 + 0.7243 TM4 +

0.0840 TM5 – 0.1800 TM7

• Wetness = 0.1509 TM1 + 0.1973 TM2 + 0.3279 TM3 + 0.3406 TM4 –

0.7112 TM5 – 0.4572 TM7

• Haze = –0.8242 TM1 + 0.0849 TM2 + 0.4392 TM3 – 0.0580 TM4 + 0.2012 TM5
– 0.2768 TM7
 IMAGE
TRANSFORMATION

IMAGE AIRTHMATIC PCA TASSELED CAP COLOUR SPACE FOURIER

IMAGE FUSION
OPERATIONS TRANSFORMATION TRANSFORMATION TRANSFORMATION TRANSFORMATION
 Color Characteristics

1. Brightness
The amount of intensity,
Brightness refers to how
much light appears to
shine from something.
2. HUE

The colours on the outermost perimeter of the colour circle are hues, which are colour in their purest
form.
3. SATURATION

Saturation refers to the intensity of colour in an image. As the saturation increases the colour appears to
be more pure. As the saturation decreases the colour appears to washed out.
Some Important Colour models
• RGB(Red, Green, Blue )Model for color monitors and video camera.
• HSI(Hue,saturation, intensity) model used for image processing
applications.
• CMY(Cyan,Magenta,Yellow )model for color printing.
RGB COLOUR MODEL
• In the RGB model, an image consists of three independent image
planes, one in each of the primary colours: red, green and blue.
Figure shows the geometry of the RGB colour model for
specifying colours using a Cartesian coordinate system.
The greyscale spectrum, i.e. those colours made
from equal amounts of each primary, lies on the
line joining the black and white vertices.
We can represent each R, G,B from 0 to 255
R+G+B=
COLOUR
IMAGE
PIXEL DEPTH
• The total number of bits used to represent each pixel in RGB image is
called pixel depth.
CMY
MODEL
CYAN MAGENTA AND
YELLOW MODEL
RGB TO CMY CONVERSION
• RGB (98, 35, 146) TO CMY
• 1ST STEP – NORMALIZATION
• R’ = 98/255 =0.384
• G’ = 35/255 =0.137
• B’ = 146/255 = 0.572
• TO CMY
• C = 1- 0.384 = 0.616
• M= 1- 0.137 = 0.863
• Y = 1- 0.572 = 0.428
HSI COLOUR
MODEL
• HUE, SATURATION AND
INTENSITY
• HUE – DESCRIBES describes the
color in the form of an angle
between [0,360] degrees.
• SATURATION - describes how
much the color is diluted with
white light. The range of the S
varies between [0,1].
• The Intensity range is between
[0,1] and 0 means black, 1
means white
 RGB TO HSI
CONVERSION

NOTE :- ALL R,G,B VALUES WILL BE USED AFTER NORMALIZATION.

 IMAGE
TRANSFORMATION

IMAGE AIRTHMATIC PCA TASSELED CAP COLOUR SPACE FOURIER

IMAGE FUSION
OPERATIONS TRANSFORMATION TRANSFORMATION TRANSFORMATION TRANSFORMATION
 Image Fusion
Image fusion is a technique of increasing the spatial resolution of a multispectral image using a
co-registered higher resolution panchromatic image of same area and time.

Multispectral image Panchromatic image

Fused image
Image Fusion

• Common techniques:
• Multiplicative fusion
• Brovey transform fusion
• PCA transform fusion
• IHS transform fusion
 • DNMSn × DNPAN = DNFMSn

Multiplicative • Where,
• DNMSn = pixel value in nth band of multi-
Fusion spectral data, DNPAN = pixel value in PAN data,
• DNFMSn = pixel value in nth band of fused multi-
spectral data.
 • Red band of fused image:
• [DNband 3 / (DNband 1 + DNband 2 + DNband 3)] × DNPAN

Brovey Transform • Green band of fused image:

Fusion • [DNband 2 / (DNband 1 + DNband 2 + DNband 3)] × DNPAN

• Blue band of fused image:

• [DNband 1 / (DNband 1 + DNband 2 + DNband 3)] × DNPAN
 • Forward PCA transform is applied to generate PCs.
• PC-1 is removed, and its numerical range (min to max) is
PCATransform determined.
• The panchromatic image is remapped so that its histogram
Fusion shape is kept constant, but it is in the same numerical range as PC-1.
• It is then substituted for PC-1.
• Reverse PCA transform is
applied.
 • Three bands (RGB) of the lower spatial resolution data are
transformed to the IHS space.
IHSTransform • A panchromatic image replaces the intensity component.
Fusion • Hue-saturation components are over-sampled to
panchromatic resolution. Re-transformed back to the
original RGB space.
Wavelet Transform Fusion
• Similar to Fourier Transform.