Digital Image Processing

Chapter 5: Image Restoration

ANISHA M. LAL
A Model of the Image
Degradation/Restoration Process
 Degradation
 Degradation function H
 Additive noise  ( x, y )
 Spatial domain

g ( x, y)  h( x, y) * f ( x, y)   ( x, y)
 Frequency domain

G(u, v)  H (u, v) F (u, v)  N (u, v)

 Restoration

g ( x, y)  Restoration Filter  fˆ ( x, y)
Noise Models
 Sources of noise
 Image acquisition, digitization,
transmission
 White noise
 The Fourier spectrum of noise is
constant
 Assuming
 Noise is independent of spatial
coordinates
 Noise is uncorrelated with respect to
the image itself
 Gaussian noise
 The PDF of a Gaussian random variable,
z,
1  ( z   ) 2 / 2 2
p( z )  e
2 

 Mean: 
 Standard deviation: 
 Variance:  2
 70% of its values will be in the range

(    ), (    )
 95% of its values will be in the range

(   2 ), (   2 )
 Rayleigh noise
 The PDF of Rayleigh noise,
2
 ( z  a )e ( z  a ) 2 / b
for z  a
p( z )   b

 0 for z  a
 Mean:   a  b/4
b( 4   )
 Variance:  2

4
 Erlang (Gamma) noise
 The PDF of Erlang noise, a0 , b
is a positive integer,
 a b z b 1 a z
 e for z  0
p( z )   (b  1)!

 0 for z  0
b
 Mean: 
a
b
 Variance:   2
2

a
 Exponential noise
 The PDF of exponential noise, a0 ,

ae a z for z  0
p( z )  
 0 for z  0
1
 Mean: 
a
1
 Variance:   2
2

a
 Uniform noise
 The PDF of uniform noise,

 1
 if a  z  b
p( z )   b  a

 0 otherwise
ab
 Mean: 
2
 Variance: 
(b  a ) 2
2

12
 Impulse (salt-and-pepper) noise
 The PDF of (bipolar) impulse noise,

 Pa for z  a

p( z )   Pb for z  b
0
 otherwise
 b  a : gray-level b will appear as a
light dot, while level a will appear like
a dark dot
 Unipolar: either Pa or Pb is zero
 Usually, for an 8-bit image, a =0
(black) and b =0 (white)
 Modeling
 Gaussian
 Electronic circuit noise, sensor noise due
to poor illumination and/or high
temperature
 Rayleigh
 Range imaging
 Exponential and gamma
 Laser imaging
 Impulse
 Quick transients, such as faulty switching
 Uniform
 Least descriptive
 Basis for numerous random number
generators
 Periodic noise
 Arises typically from electrical or
electromechanical interference
 Reduced significantly via frequency
domain filtering
Restoration in the Presence of Noise
Only-Spatial Filtering

 Degradation
 Spatial domain
g ( x, y)  f ( x, y)   ( x, y)
 Frequency domain

G(u, v)  F (u, v)  N (u, v)

 Mean filters
 Arithmetic mean filter

ˆf ( x, y )  1
 g ( s, t )
mn ( s ,t )S xy
 Geometric mean filter
1
  mn
fˆ ( x, y )    g ( s, t )
( s ,t )S xy 
 Harmonic mean filter
 Works well for salt noise, but fails fpr
pepper noise

mn
fˆ ( x, y ) 
1

( s ,t )S xy g ( s, t )
 Contraharmonic mean filter
 Q0 : eliminates pepper noise
 Q0 : eliminates salt noise

 g ( s ,
( s ,t )S xy
t ) Q 1

fˆ ( x, y ) 
 g ( s, t )
( s ,t )S xy
Q
 Usage
 Arithmetic and geometric mean filters:
suited for Gaussian or uniform noise
 Contraharmonic filters: suited for
impulse noise
 Order-statistics filters
 Median filter
 Effective in the presence of both bipolar
and unipolar impulse noise

fˆ ( x, y )  median{g ( s, t )}
( s ,t )S xy
 Max and min filters
 max filters reduce pepper noise

fˆ ( x, y )  max {g ( s, t )}
( s ,t )S xy

 min filters salt noise

fˆ ( x, y )  min {g ( s, t )}
( s ,t )S xy
 Midpoint filter
 Works best for randomly distributed noise,
like Gaussian or uniform noise

ˆf ( x, y )  1  max {g ( s, t )}  min {g ( s, t )}

2 ( s ,t )S xy ( s ,t )S xy 
 Alpha-trimmed mean filter
 Delete the d/2 lowest and the d/2 highest
gray-level values
 Useful in situations involving multiple
types of noise, such as a combination of
salt-and-pepper and Gaussian noise

1
fˆ ( x, y ) 
mn  d
 g ( s, t )
( s ,t )S xy
r