A Lecture on

Introduction to

Image Restoration
“To introduce the basic concept of Image
Restoration in Digital Image Processing”
Outlines

 What is Image Restoration.  Some Basics Filter

 Image Enhancement vs.  Advanced Image Restoration.

Image Restoration.  Conclusions.

 Image Degradation Model.  Tools for DIP.

 Applications.
 Noise Models.

 Estimation of Degradation

Model.

 Restoration Techniques.
Lets start

 What is Image Restoration.  Some Basics Filter

 Image Enhancement vs.  Advanced Image Restoration.

Image Restoration.  Conclusions.

 Image Degradation Model.  Tools for DIP.

 Applications.
 Noise Models.

 Estimation of Degradation

Model.

 Restoration Techniques.
What is Image Restoration.
What is Image Restoration?

 Image restoration attempts to restore images that have been degraded

 Identify the degradation process and attempt to reverse it.

 Almost Similar to image enhancement, but more objective.

Fig: Degraded image Fig: Restored image

Where we reached?

 What is Image Restoration.  Some Basics Filter

 Image Enhancement vs.  Advanced Image Restoration.

Image Restoration.  Conclusions.

 Image Degradation Model.  Tools for DIP.

 Applications.
 Noise Models.

 Estimation of Degradation

Model.

 Restoration Techniques.
Image enhancement vs. Image Restoration

• Image restoration assumes a degradation model that is known or can be

estimated.
• Original content and quality does not mean Good looking or appearance.
• Image Enhancement is subjective, where as image restoration is objective
process.
• Image restoration try to recover original image from degraded with prior
knowledge of degradation process.
• Restoration involves modeling of degradation and applying the inverse
process in order to recover the original image.
• Although the restore image is not the original image, its approximation of
actual image.
Where we reached?

 What is Image Restoration.  Some Basics Filter

 Image Enhancement vs.  Advanced Image Restoration.

Image Restoration.  Conclusions.

 Image Degradation Model.  Tools for DIP.

 Applications.
 Noise Models.

 Estimation of Degradation

Model.

 Restoration Techniques.
Degradation Model?

 Objective: To restore a degraded/distorted image to its original content

and quality.
g(x,y)
Degradation ^
f(x,y) Restoration f(x,y)
Function h Filters

Degradation Restoration
ŋ(x,y)
 Spatial Domain: g(x,y)=h(x,y)*f(x,y)+ ŋ(x,y)
 Frequency Domain: G(u,v)=H(u,v)F(u,v)+ ŋ(u,v)
 Matrix: G=HF+ŋ
Going On….!

 What is Image Restoration.  Some Basics Filter

 Image Enhancement vs.  Advanced Image Restoration.

Image Restoration.  Conclusions.

 Image Degradation Model.  Tools for DIP.

 Applications.
 Noise Models.

 Estimation of Degradation

Model.

 Restoration Techniques.
Noise Models and Their PDF

• Different models for the image

Rayleigh
noise term η(x, y) Gaussian

 Gaussian
 Most common model
 Rayleigh Erlang Exponential
 Erlang or Gamma
 Exponential
 Uniform
 Impulse Uniform Impulse
 Salt and pepper noise
Noise Models Effects

Histogram to go here

Fig: Original Image Fig: Original Image histogram

Noise Models Effects contd1…
Noise Models Effects contd2…
Going On….!

 What is Image Restoration.  Some Basics Filter

 Image Enhancement vs.  Advanced Image Restoration.

Image Restoration.  Conclusions.

 Image Degradation Model.  Tools for DIP.

 Applications.
 Noise Models.

 Estimation of Degradation

Model.

 Restoration Techniques.
Estimation of Degradation Model.

 Weather the spatial or frequency domain or Matrix, in all cases knowledge

of degradation function is important.

 Estimation of H is important in image restoration.

 There are mainly three ways to estimate the H as follows-

 By Observation

 By Experimentation.

 Mathematical Modeling

• After approximation the degradation function, we apply the BLIND

CONVOLUTION to restore the original image.

Observation

 No knowledge of degraded function is given.

 Observing on g(x,y), try to estimate the degraded function in the region

which have simpler structure.

gs(x,y) Gs(u,v)
Hs(u,v)= Gs(u,v)/ Fs(u,v)
fs(x,y) Fs(u,v)
Observation contd…

gs(x,y)

Gs(u,v)

Hs(u,v)= Gs(u,v)/ Fs(u,v)

Fs(u,v)

fs(x,y)
Experimentation

• Try to imaging set-up similar to original.

 Impulse response and impulse simulation.

 Objective to find H which have similar result of degradation as original

one.
Impulse Impulse Response

Fig: Impulse Simulation

Experimentation contd…

 Here f(x,y) is impulse.

 F(u,v)=>A (a constant).

 G(u,v)=H(u,v)F(u,v).

 H(u,v)=G(u,v)/A.

 Objective is training and testing.

Note: The intensity of impulse is very high, otherwise noise can dominate to
impulse.
Mathematical Modeling

 If you have the mathematical model, you have inside the degradation

process.

 Atmospheric turbulence can be possible to mapping in mathematical model.

 One e.g. of mathematical model

 k gives the nature of turbulence .

Mathematical Modeling contd..
Atmospheric Turbulence blur examples

Fig: Negligible Turbulence Fig: Severe Turbulence, k=0.0025

Fig: Mid Turbulence, k=0.001 Fig: Low Turbulence, k=0.00025

Present Position

 What is Image Restoration.  Some Basics Filter

 Image Enhancement vs.  Advanced Image Restoration.

Image Restoration.  Conclusions.

 Image Degradation Model.  Tools for DIP.

 Applications.
 Noise Models.

 Estimation of Degradation

Model.

 Restoration Techniques.
Restoration Techniques.

 Inverse Filtering.

 Minimum Mean Squares Errors.

 Weiner Filtering.

 Constrained Least Square Filter.

 Non linear filtering

 Advanced Restoration Technique.

Filter used for Restoration Process

 Mean filters
 Adaptive filters
 Arithmetic mean filter
 Adaptive local noise reduction
 Geometric mean filter filter.
 Harmonic mean filter  Adaptive median filter
 Contra-harmonic mean filter

 Order statistics filters

 Median filter

 Max and min filters

 Mid-point filter

 alpha-trimmed filters
Filtering to Remove Noise-AMF

 Use spatial filters of different kinds to remove different kinds of noise

 Arithmetic Mean :

ˆf ( x, y )  1
 g ( s, t )
mn ( s ,t )S xy
 This is implemented as the simple smoothing filter Blurs the image to
remove noise.

1/ 1/ 1/
9 9 9
1/ 1/ 1/
9 9 9
1/ 1/ 1/
9 9 9
Filtering to Remove Noise-GMF

 Geometric Mean:

1
  mn
fˆ ( x, y )    g ( s, t )
( s ,t )S xy 
 Achieves similar smoothing to the arithmetic mean, but tends to lose less

image detail.
Filtering to Remove Noise-HMF

 Harmonic Mean:

mn
fˆ ( x, y ) 
1

( s ,t )S xy g ( s, t )
 Works well for salt noise, but fails for pepper noise

 Satisfactory result in other kinds of noise such as Gaussian noise

Filtering to Remove Noise-CHMF

 Contra-harmonic Mean:

 g ( s, t )
( s ,t )S xy
Q 1

fˆ ( x, y ) 
 g ( s ,
( s ,t )S xy
t ) Q

 Q is the order of the filter and adjusting its value changes the filter’s
behaviour.

 Positive values of Q eliminate pepper noise.

 Negative values of Q eliminate salt noise.
Result of AMF and GMF

Fig: Original Image Fig: Gaussian Noise

Fig: Result of 3*3 AM Fig: Result of 3*3 GM

Result of Contra-harmonic Mean Filter

Fig: Original Image with Pepper noise Fig: After filter by 3*3 CHF, Q=1.5

Fig: Original Image with Salt noise Fig: After filter by 3*3 CHF, Q=-1.5
Beware: Q value in Contra-harmonic Filter

 Choosing the wrong value for Q when using the contra-harmonic filter can

have drastic results.

Order Statistics Filters

 Spatial filters that are based on ordering the pixel values that make up

the neighbourhood operated on by the filter

 Useful spatial filters include

 Median filter.

 Maximum and Minimum filter.

 Midpoint filter.

 Alpha trimmed mean filter.

Median Filter

 Median Filter:

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

 Excellent at noise removal, without the smoothing effects that can occur

with other smoothing filters

 Best result for removing salt and pepper noise.

Maximum and Minimum Filter

 Max Filter:

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

 Min Filter:

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

 Max filter is good for pepper noise and min is good for salt noise
Midpoint Filter

 Midpoint Filter:

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

2 ( s ,t )S xy ( s ,t )S xy 
 Good for random Gaussian and uniform noise
Alpha-Trimmed Mean Filter

 Alpha-Trimmed Mean Filter:

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

 Here deleted the d/2 lowest and d/2 highest grey levels, so gr(s, t)
represents the remaining mn – d pixels
Result of Median Filter

Fig 1: Salt & Pepper noise Fig2: Result of 1 pass Med 3*3

Fig3: Result of 2 pass Med 3*3 Fig4: Result of 3 pass Med 3*3
Result of Max and Min Filter

Fig: Corrupted by Pepper Noise Fig: Corrupted by Salt Noise

Fig: Filtering Above,3*3 Max Filter Fig: Filtering Above,3*3 Min Filter
 Periodic Noise

 Typically arises due to electrical or electromagnetic interference.

 Gives rise to regular noise patterns in an image

 Frequency domain techniques in the Fourier domain are most effective at

removing periodic noise

Fig: periodic Noise

 Band Reject Filters

 Removing periodic noise form an image involves removing a particular range of

frequencies from that image.

 Band reject filters can be used for this purpose.

 An ideal band reject filter is given as follows:

 W
1 if D(u, v)  D0  2
 W W
H (u, v)  0 if D0   D(u, v)  D0 
 2 2
1 if D(u, v)  D0  W
 2
Band Reject Filters contd..

 The ideal band reject filter is shown below, along with Butterworth

and Gaussian versions of the filter.

Ideal Band Butterworth Gaussian

Reject Filter Band Reject Band Reject
Filter (of order 1) Filter
Result of Band Reject Filter

Fig: Corrupted by Sinusoidal Noise Fig: Fourier spectrum of Corrupted Image

Fig: Butterworth Band Reject Filter Fig :Filtered image

 Conclusions-What we learnt…

 Restore the original image from degraded image, if u have clue about

degradation function, is called image restoration.

 The main objective should be estimate the degradation function.

 If you are able to estimate the H, then follow the inverse of degradation

process of an image.

 Whether spatial or frequency domain.

 Spatial domain techniques are particularly useful for removing random

noise.

 Frequency domain techniques are particularly useful for removing periodic

noise.
Advanced Image Restoration
• Adaptive Processing
 Spatial adaptive
 Frequency adaptive

• Nonlinear Processing
 Thresholding, coring …
 Iterative restoration

• Advanced Transformation / Modeling

 Advanced image transforms, e.g., wavelet …
 Statistical image modeling

• Blind Deblurring or Deconvolution

Lets conclude..!

 What is Image Restoration.  Some Basics Filter

 Image Enhancement vs.  Advanced Image Restoration.

Image Restoration.  Conclusions.

 Image Degradation Model.  Tools for DIP.

 Applications.
 Noise Models.

 Estimation of Degradation

Model.

 Restoration Techniques.
 Conclusions-What we learnt…

 Restore the original image from degraded image, if u have clue about

degradation function is called image restoration.

 The main objective should be estimate the degradation function.

 If you are able to estimate the H, then follow the inverse of degradation

process of an image.

 Weather spatial or frequency domain.

 Spatial domain techniques are particularly useful for removing random

noise.

 Frequency domain techniques are particularly useful for removing periodic

noise.
Popular Image Processing Software Tools

 CVIP tools

(Computer Vision and Image Processing tools)

 Intel Open Computer Vision Library

 Microsoft Vision SDL Library

 MATLAB

 KHOROS
 Applications of
Digital Image Processing?
Applications of Digital Image Processing

 Identification.

 Computer Vision or Robot vision.

 Steganography.

 Image Enhancement.

 Image Analysis in Medical.

 Morphological Image Analysis.

 Space Image Analysis.

 Bottling and IC Industry……….etc.