Scribd
Upload
19 views29 pages

Unit 3 - 1 - Image - Restoration

The document discusses image processing, specifically focusing on image restoration and enhancement. It explains the differences between enhancement, which improves the appearance of images, and restoration, which aims to recover original image quality from degradation. Various noise models and filtering techniques, including mean filters and median filters, are described for effective image restoration in the presence of noise.

Uploaded by

privateprivacymail
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
19 views29 pages

Unit 3 - 1 - Image - Restoration

The document discusses image processing, specifically focusing on image restoration and enhancement. It explains the differences between enhancement, which improves the appearance of images, and restoration, which aims to recover original image quality from degradation. Various noise models and filtering techniques, including mean filters and median filters, are described for effective image restoration in the presence of noise.

Uploaded by

privateprivacymail
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 29

EC 703:

IMAGE PROCESSING AND


COMPUTER VISION

By:
1
Rahul K Patel
ECE Department
I.I.I.T, Surat
IMAGE RESTORATION
ENHANCEMENT V/S. RESTORATION
 Image Enhancement:
 A process which aims to improve bad

 Images so they will “look” better.

 Image Restoration

Restoration attempts to recover an image that has


been degraded by using a priori knowledge of
the degradation phenomenon. Thus, estoration
techniques are oriented toward modeling the
degradation and applying the inverse process in
order to recover the original image.
MODEL OF IMAGE DEGRADATION AND
RESTORATION
 The purpose of image restoration is to restore a
degraded/distorted image to its original content and quality.
 Distinctions to Image Enhancement
 Image restoration assumes a degradation model that is
known or can be estimated.
 Original content and quality ≠ Good looking
 if H is a linear, position-invariant operator, then
the degraded image is given in the spatial
domain by

where h(x, y) is the spatial representation of


the degradation function.
 in the frequency domain
NOISE MODELS
 Most types of noise are
modeled as known probability
density functions
 Noise model is decided based
on understanding of the physics
of the sources of noise.
 Gaussian: poor illumination
 Rayleigh: range image
 Gamma, exp: laser imaging
 Impulse: faulty switch during
imaging,
 Uniform is least used.
 Parameters can be estimated
based on histogram on small
flat area of an image
RESTORATION IN THE PRESENCE
OF NOISE ONLY-SPATIAL FILTERING

 g(x,y)=f(x,y)+n(x,y);

 G(u,v)=F(u,v)+N(u,v);
ARITHMATIC MEAN FILTER

f’(x,y)=1/mn* Σ g(s,t)
(s,t) єSXY

Noise is reduced as a result of


blurring
GEOMATRIC MEAN FILTER

f’(x,y)= [Π g(s,t)] 1/mn


(s,t) єSXY

Geometric mean filter achieves


smoothing comparable to the
arithmetic mean filter, but it tends to
loss less image detail in the process.
HARMONIC MEAN FILTER

f’(x,y)=mn/ [Σ 1/g(s,t)]
(s,t) єSXY

The harmonic mean filter works well for


salt noise, but fails for pepper noise. It
does well also with other types of noise
like Gaussian noise.
CONTRAHARMONIC MEAN

 f’(x,y)= [Σ g(s,t)] Q+1


(s,t) єSXY
--------------------------------

[Σ g(s,t)] Q
(s,t) єSXY

 Where Q is called order of the filter


 This filter is well suited for reducing or virtually
eliminating the effects of salt and peppers noise.
 Q=0 it becomes arithmetic mean, and Q = -1
become harmonic mean.
Selection of sign of Q depend on the noise type,
noise is dark or light.
If wrong sign of Q is selected it gives improper results
 Order Statistics filter
MEDIAN FILTER

• f’(x,y)=median{g(s,t)}
(s,t) єSXY

• For bipolar and unipolar impulse


noise (salt and pepper noise) median
filter provide excellent noise-
reduction capabilities, with
considerably less blurring than linear
smoothing filters of similar size.
MAX AND MIN FILTERS
 f’(x,y)=max{g(s,t)}
(s,t) єSXY

 This filter is useful for finding the brightest


point in an image. Reducing peeper noise.
 f’(x,y)=min{g(s,t)}
(s,t) єSXY

 This filter is used for finding the darkest point


in an image. It reduces salt noise.
MIDPOINT FILTER

 f’(x,y)=1/2*[max {g(s,t)} +min{g(s,t)} ]


 (s,t) єS (s,t) єS

 This filter combines order statistics and averaging.


 This filter works best for randomly distributed noise,
like Gaussian and uniform noise.
ALPHA-TRIMMED MEAN FILTER
 Suppose that we delete the d/2 lowest and d/2
highest gray-level values of g(s,t) in the
neighborhood.
 f’(x,y)=1/(mn-d)*Σ gr(s,t)
( s,t) єSXY

d can range from 0 to mn-1. d=0, the filter


reduces to arithmetic mean filter. If we choose
d=(mn-1)/2, the filter becomes the median
filter.
 It is useful in situations involving multiple
types of noise, such as a combination of salt-
and-pepper and Gaussian noise.

You might also like