BORANA UNIVERSITY
COLLEGE OF ENGINEERING AND
TECHNOLOGY
DEPARTMENT OF COMPUTER SCIENCE
Course Title: Computer Vision and Image processing
Course instructor: Guyita G.
Target Group: 4th Year First semester
Chapter Five:
Image Restoration and Reconstruction
Outline of the Chapter
ØIntroduction
ØModels of images and noise
ØEstimation of noise parameters
ØRestoration in the presence of noise
1/27/2025 2
Introduction
ØThe goal of restoration techniques is to improve an image
in some predefined sense in image enhancement.
ØRestoration attempts to recover an image that has been
degraded by using a priori knowledge of the degradation
phenomenon.
ØRestoration techniques are oriented toward modeling the
degradation and applying the inverse process in order to
recover the original image.
ØImage Restoration refers to a class of methods that aim to
remove or reduce the degradations that have occurred
while the digital image was being obtained.
1/27/2025 3
Model of the Image Degradation/Restoration
Process
ØD e g r a d a t i o n p r o c e s s o p e r a t e s o n a
degradation function that operates on an
input image with an additive noise term.
ØModel image degradation as an operator ℋ
that, together with an additive noise term,
operates on an input image f (x,y) to produce
a degraded image g(x,y).
ØGiven g(x,y ), some knowledge about ℋ, and
some knowledge about the additive noise term
ɳ(x,y), the objective of restoration is to obtain
an estimate f ˆ(x,y) of the original image.
ØWe want the estimate to be as close as
possible to the original image and, in general,
the more we know about ℋ and ɳ, the closer f
ˆ(x,y)
1/27/2025
will be to f (x,y). 4
Cont.…
ØIf 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.
ØIn frequency domain we may write this equation
as
G(u,v)=F(u,v)H(u,v)+N(u,v)
ØThe terms in the capital letters are the Fourier
Transform of the corresponding terms in the
spatial domain.
1/27/2025 5
Cont.…
A model of the image Degradation / Restoration Process
1/27/2025 6
Cont.…
1/27/2025 7
Noise Models
ØThe principal source of noise in digital images arises during
image acquisition and /or transmission.
ØThe performance of imaging sensors is affected by a variety of
factors, such as environmental conditions during image
acquisition and by the quality of the sensing elements
themselves.
ØImages are corrupted during transmission principally due to
interference in the channels used for transmission.
ØSince main sources of noise presented in digital images are
resulted from atmospheric disturbance and
Øimage sensor circuitry, following assumptions can be made i.e.
the noise model is spatial invariant (independent of spatial
location).
ØThe noise model is uncorrelated with the object function.
ØFor example, an image transmitted using a wireless network
might be corrupted by lightning or other atmospheric
disturbance.
1/27/2025 8
Cont.…
Some important noise probability density functions (PDFs)
The most common noise PDFs found in image processing
applications are:
ØGaussian Noise
ØRayleigh Noise
ØExponential Noise
ØUniform Noise
ØSalt-and-Pepper Noise
ØPeriodic Noise
1/27/2025 9
Gaussian Noise
ØBecause of its mathematical tractability in both the
spatial and frequency domains, Gaussian noise models
are used frequently in practice.
ØThe PDF of a Gaussian random variable, z, is defined by
the following familiar expression:
1/27/2025 10
Cont.…
where z represents intensity, z’ is the mean
(average) value of z, and δ is its standard
deviation.
1/27/2025 11
Rayleigh Noise
The PDF of Rayleigh noise is given by
The mean and variance of z when this random
variable is characterized by a Rayleigh PDF are
1/27/2025 12
Cont.…
ØThe displacement from the origin, and the fact that the
basic shape of the density is skewed to the right.
ØThe Rayleigh density can be quite useful for modeling the
shape of skewed histograms.
1/27/2025 13
Erlang (Gamma) Noise
The PDF of Erlang noise is
where the parameters are such that a>b , b is a positive
integer, and “!” indicates factorial. The mean and variance
of z are
and
1/27/2025 14
Cont.…
The plot of this density.
1/27/2025 15
Exponential Noise
The PDF of exponential noise is given by
where a > 0. The mean and variance of z
are
and
1/27/2025 16
Cont.…
The plot of the exponential density function.
1/27/2025 17
Uniform Noise
The PDF of uniform noise is
The mean and variance of z are
and
1/27/2025 18
Cont.…
The plot of the uniform density
1/27/2025 19
Impulse (Salt-and-Pepper) Noise
ØThe noise is signal dependent, and is multiplied to the
image.
ØIf k represents the number of bits used to represent the
intensity values in a digital image, then the range of
possible intensity values for that image is [0, 2k-1] (e.g.,
[0,255] for an 8-bit image).
ØThe PDF of salt-and-pepper noise is given by
1/27/2025 20
Periodic Noise
ØPeriodic noise in images typically arises from
electrical or electromechanical interference
during image acquisition.
ØThis is the only type of spatially dependent
noise
Øperiodic noise can be reduced significantly via
frequency domain filtering.
1/27/2025 21
Estimation of noise parameters
ØThe parameters of periodic noise typically are
estimated by inspection of the Fourier spectrum.
ØPeriodic noise tends to produce frequency spikes
that often can be detected even by visual
analysis.
ØAnother approach is to attempt to infer the
periodicity of noise components directly from
the image,
Øbut this is possible only in simplistic cases.
ØAutomated analysis is possible in situations in
which the noise spikes are either exceptionally
pronounced, or when knowledge is available
about the general location of the frequency
components of the interference.
1/27/2025 22
Cont.…
ØThe parameters of noise PDFs may be known
partially from sensor specifications, but it is
often necessary to estimate them for a
particular imaging arrangement.
ØIf the imaging system is available, one simple
way to study the characteristics of system
noise is to capture a set of “flat” images.
ØFor example, in the case of an optical sensor,
this is as simple as imaging a solid gray board
that is illuminated uniformly.
ØThe resulting images typically are good
indicators of system noise.
1/27/2025 23
Restoration in the presence of noise
ØWhen an image is degraded only by additive noise,
and
ØThe noise terms generally are unknown, so subtracting
them from g(x,y) [G(U,V)] to obtain f(x,y) [F(U,V)]
typically is not an option.
ØIn the case of periodic noise, sometimes it is possible to
estimate N(u,v) from the spectrum of G(u,v)
ØIn this case N(u,v) can be subtracted from G(u,v) to
obtain an estimate of the original image, but this type of
knowledge is the exception, rather than the rule.
1/27/2025 24
Cont.…
ØSpatial filtering is the method of choice for estimating f(x,y)
[i.e., denoising image g(x,y)] in situations when only additive
random noise is present.
ØThere are some noise reduction mechanisms:
Mean Filter:
§ Arithmetic mean filter
§ Geometric mean filter
§ Contraharmonic mean filter
Order static filters
§ Median filter
§ Max and Min filter
§ Mid point filter
§ Alpha
1/27/2025
Trimmed mean filter 25
The End
Of
The
Unit
1/27/2025 26