Image Enhancement (Spatial Domain Methods)

What Is Image Enhancement?

Image enhancement is the process of making images more useful The reasons for doing this include:

Highlighting interesting detail in images

Removing noise from images Making images more visually appealing

Image Enhancement

Enhance otherwise hidden information Filter important image features Discard unimportant image features Emphasize, sharpen or smoothen image features

Classification of Image enhancement

Spatial Domain Process intensity of pixels Two types- intensity transformation and spatial filtering

Transform Domain
Transform image, process it and then find inverse transform to get image in spatial domain

Basic of Spatial Domain Filtering

g (x, y) = T[ f (x, y)] f (x, y) is the input image g (x, y) is the processed image
(x, y) Origin y

and T is operator defined over some neighbourhood of (x, y)

x Image f (x, y)

Point Processing

Point processing operations take the form s=T(r)

s refers to the processed image pixel value and r

refers to the original image pixel value

T is referred to as a grey level transformation function

or a point processing operation f(x,y) g(x,y)

Spatial Domain

The operator T can be defined over

The set of pixels (x,y) of the image The set of neighborhoods, N(x,y) of
each pixel

Classification of spatial domain

Point operation

Mask operation
Global operation

Point operation
Brightness modification
Contrast manipulation

Histogram manipulation

Spatial Domain

Operation on the set of image-pixels

0 10

3 6

0 5

12 200 20

100 10

(Operator: Div. by 2)

Spatial Domain

Operation on the set of neighborhood

6 8 (Operator: sum)

12 200
6 8 2 0 10 226

12 200 20

Spatial Domain

Global Operation
6 8 2 0

12 200 20 10
(Operator: sum) 11 13 3 0 14

14 220 23 5 2 5 20 1 3 0 4

Transformations

Gray Level/Intensity Transformations

Brightness modification Image negatives Piecewise-Linear transformation

Functions Log transformations Power Law transformations

Intensity Level Transformations

Linear

Negative/Identity

Logarithmic

Log/Inverse log

Power

law

nth power/nth root

Image Negative
Suited for enhancing white or grey detail embedded in dark region and black area predominates

Output gray level

g(x,y)= 255- f(x,y)

Input gray level

Logarithmic Transformations
The log transformation maps a narrow range of low input grey level values into a wider range of output values

The inverse log transformation performs the opposite transformation

g(x,y) = c * log (1+ f(x,y))

Log Transformations
Input grey level values has large range of values

InvLog

Log

Log Transformations

Logarithm of FT reveals more details Range, 0 to 106 becomes 0 to 6.2

Power Law Transformations T(f) = c*f

Transformations

>1 Compresses dark values Expands bright values <1 Expands dark values Compresses bright values

Power Law Transformations

s=c*r Map a narrow range of dark input values into a wider range of output values or vice versa

Varying gives a whole family of curves

Application of gamma correction

A cathode ray tube (CRT) converts a video signal to light in a nonlinear way. The light intensity I is proportional to a power () of the source voltage V, (I=V) For a computer CRT, is about 2.2 To view image on monitors -correction is required

Application of gamma-correction

Power Law Example

Magnetic Resonance (MR) image of a fractured human spine

Power Law Example ( = 0.6)

= 0.6
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 Old Intensities
Transformed Intensities

Power Law Example ( = 0.4)

= 0.4
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 Original Intensities

Transformed Intensities

Power Law Example ( = 0.3)

= 0.3
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 Original Intensities

Transformed Intensities

Power Law Example (cont)

s = r 0.6

s = r 0.4

Power Law Example (Image with washed out appearance)

An aerial view of a runway

Image after gamma correction ( > 1)

= 5.0
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 Original Intensities

Transformed Intensities

s = r 3.0

Different curves highlight different detail

s = r 4.0

Brightness/contrast modification

g(m,n) = f(m,n) + k (increase brightness) g(m,n) = f(m,n) k (decrease brightness)

Piecewise Linear Transformations

Thresholding Function

g(x,y) = L-1, f(x,y) > t = 0, f(x,y) < t t = threshold level

Output gray level Input gray level

Thresholding
Thresholding transformations are particularly useful for segmentation in which we want to isolate an object of interest from a background

1.0 r > threshold s= 0.0 r <= threshold

Contrast stretching

Gray/Intensity Level Slicing

Highlight a specific range of gray values Two approaches:

Display high value for range of interest, low value else (discard background) Display high value for range of interest, original value else (preserve background)

Gray Level Slicing, example

Bit Plane Slicing

Isolate particular bits of intensity value

Shows contribution of each bit

Higher-order bits usually contain most of the significant visual information

Lower-order bits

contain subtle details

Intensity = (b7 b6 b5 b4 b3 b2 b1b0)

BP 0 BP 5 BP 7

Bit Plane Slicing (example)

Intensity = (b7 b6 b5 b4 b3 b2 b1b0)

Bit Plane Slicing (plane 1)

Bit Plane Slicing (plane 2)

Bit Plane Slicing (plane 3)

Bit Plane Slicing (plane 4)

Bit Plane Slicing (plane 5)

Bit Plane Slicing (plane 6)

Bit Plane Slicing (plane 7)

Bit Plane Slicing (plane 8)

Bit Plane Slicing (cont)

Reconstructed image using only bit planes 8 and 7 Reconstructed image using only bit planes 8, 7 and 6 Reconstructed image using only bit planes 7, 6 and 5

Histogram
Plot of number of occurrences of grey levels against grey level values
0 1 1 3 2 0 3 1 4 1 5 2

Number of Pixels

gray level

Histogram of image

Histogram Examples

Histogram Examples (cont)

Histogram Examples (cont)

Histogram Examples (cont)

Histogram Examples (cont)

High contrast image has the most evenly spaced histogram

Histogram Equalization

Equal number of pixels for every gray-value Histogram is constant Preprocessing technique to enhance contrast in natural images Find gray level transformation function T to transform image f such that the histogram of T(f) is equalized Spreading out the frequencies in an image (or equalising the image) improves dark or washed out images

Equalisation Transformation Function

Histogram Equalisation

rk:input intensity sk:processed intensity k: the intensity range

nj:the frequency of intensity j n: the sum of all frequencies

sk T ( rk )
p j (rj )
j 1 k

j 1

nj n

Histogram Equalisation

Spread out gray levels to evenly distribute

in the range Find cumulative frequency distribution Normalize by dividing by total number of pixels Multiply by maximum gray value Map gray levels

Equalisation Transformation Functions

Histogram Equalization

How does it work ?

Mean value (or average gray level)

m = i ri p(ri) =1*0.3+2*0.1+3*0.2+4*0.1+5*0.2+6*0.1 =2.6

Mean value represents overall brightness

P(r)

0.3 0.2 0.1 0.0

1 2 3 4 5 6 r

Variance

gives a measure of the distribution of

histogram values around the mean

v = 2 = i (ri-m)2 p(ri) M=2.6, v1=(1-2.6)2x0.3+

0.3 0.2 0.1 0.0 0.3 0.2 0.1 0.0

V1 =3.34

V2=0.24

Standard Deviation
A value on the gray level axis, showing average distance of all pixels to the mean = sqrt(v)

0.3 0.2 0.1 0.0

0.3 0.2 0.1 0.0

D1

>

D2

Histograms

Variance and Standard Deviation of the histogram represent average contrast of the image The higher the Variance (=the higher the Standard Deviation), the higher the images contrast

Histograms

Histograms with mean and Standard deviation

M=0.73

D=0.32

M=0.71 D=0.27