Sei sulla pagina 1di 32


A topic
Submitted In Partial Fulfillment Of The Requirement For The Award Of
Degree Of


Submitted To


Session 2017-2018
Submitted by
Pankaj Agrawal

Under the supervision of

Asst. Prof. Mr. Rahul Satbhaiya
Department of Civil Engineering
Disaster Resistance Building
Disaster Resistance-

Concrete is resistant to wind, hurricanes, floods,

and fire. Concrete, as a structural material and as
the building exterior skin, has the ability to
withstand natures normal deteriorating
mechanisms as well as natural disasters. Properly Masonry Safe Room
designed, reinforced concrete is resistant to
earthquakes and provides blast protection for
occupants. Concrete safe rooms help provide protection from earthquakes,
tornadoes, hurricanes, fires, and other disasters.

Types of Disaster resistance-

Fire Resistance-
Concrete offers non-combustible construction that helps contain a fire
within boundaries.

As a separation wall, concrete helps to prevent a fire from spreading

throughout a building.
As an exterior wall or roof, concrete helps to prevent a fire from
jumping from building to building.
During wild fires, concrete walls and roofs help provide protection to
human life and the occupants possessions within a building.
Concrete helps contain a fire even if no water supply is available,
whereas sprinklers rely on a water source.
Concrete that endures a fire can often be reused when the building is

Tornado, Hurricane, and Wind Resistance

Concrete is resistant to tornadoes, hurricanes, and wind. Following

Hurricane Katrina, a concrete house was the sole house left standing in a
Pass Christian, MS, and neighborhood.
Investigators have learned from previous hurricanes

Asphalt shingles often failed due to holes created

by staple guns. Nails held better than staples if
they were properly placed.
Clay roof tiles resisted wind forces better than
asphalt shingles but were apt to shatter if hit by The Sundbergs' home, in
flying debris. the Pass Christian, MS, area

Concrete roof tiles suffered similar damage as clay affected by Hurricane

Katrina, is shown in the

roof tiles from debris, but were more resistant to yellow circle and is a prime
shattering than clay tiles. example of the durability of
Asphalt gravel roofs, if not well maintained, were concrete homes. (PCA
photo from FEMA)
flaked off in layers by the wind, exposing sub-
Plywood sheathing failures were due to inadequate nailing.
Particle board does not provide a good base for the attachment of
surface roofing materials.
Gables were more prone to failure than hip roofs. Gables constructed
of concrete masonry fared much better than frame construction.
Inadequate attachment to walls and inadequate lateral support caused
many failures of gables, particularly wood frame gables.
Concrete block walls performed well. Concrete masonry construction
was more forgiving of poor craftsmanship than wood frame
construction. Compliance with the SSTD 10-93, Standard for Hurricane
Resistant Residential Construction or the provisions of ACI 530/ASCE
5/TMS402-95 would have probably reduced the amount of damage
observed in these structures.
Masonry veneer also performed well when properly constructed and
connected to the structure. Damaged veneers were invariably a result
of corroded, inadequate, or improperly embedded ties. Masonry veneer
structures subjected to storm surges were able in many cases to
withstand the storm surge better than wood frame houses without
Wood frame walls performed poorly unless well designed and
Loads on building components and connections are significantly
increased when the envelope is breached by high wind or flying debris.
Masonry systems appeared to resist breaching as well, if not better,
than other wall systems.
Windows and doors need to be carefully installed. Windows must be
protected with hurricane shutters.
Flood Resistance
Concrete is not damaged by water; concrete that does not dry out
continues to gain strength in the presence of moisture. Concrete
submerged in water absorbs very small amounts of water over long
periods of time, and the concrete is not damaged. In flood-damaged
areas, concrete buildings are often salvageable. Concrete dams and
levees are used for long-lasting flood control.

In the rebuilding of New Orleans after Hurricane Katrina, architects

and engineers are looking at structures that will keep water out and
not shift or float away when submersed in floodwaters. One solution
is reinforced concrete walls to the roof height with a 12-in. thick
concrete slab. In one example, the slab will be kept in place with 8-
in. helical anchors drilled 10 to 13 feet into the ground (Architect
Hank Browne and engineers DMK Group, April 2006 Building Design
and Construction).

Concrete will only contribute to moisture problems in buildings if it

is enclosed in a system that traps moisture between the concrete
and other building materials. For instance, a vinyl wall covering in
hot and humid climates will act as a vapor retarder and moisture
can get trapped between the concrete and the wall covering. For
this reason, impermeable wall coverings (such as vinyl wallpaper)
should not be used on concrete walls.

High Humidity and Wind-Driven Rain-

Concrete is not affected by wind-driven rain and moist outdoor air in hot
and humid climates because it is impermeable to air infiltration and wind-
driven rain. Moisture that enters a building must come through joints
between concrete elements. Annual inspection and repair of joints will
minimize this potential. More importantly, if moisture does enter through
joints, it will not damage the concrete. Good practice for all types of wall
construction is to have permeable materials that breathe (are allowed to
dry) on at least one surface and to not encapsulate concrete between two
impermeable surfaces. Concrete will dry out if not covered by
impermeable treatments.
Earthquake Resistance-

Concrete is resistant to earthquakes. Earthquakes in

Guam, the United States (Richter scale 8.1); Manila,
the Philippines (Richter scale 7.2); and Kobe, Japan
(Richter scale 6.9) have subjected concrete
buildings to some of natures deadliest forces.
Concrete framing systems have a proven capacity to
In figure
withstand these major earthquakes. Another
pertinent example is the 1994 Northridge, CA, earthquake (Richter scale
6.8). It was one of the costliest natural disasters in U.S. history, with total
damages estimated at $20 billion. Most engineered structures within the
affected region performed well, including structures with concrete
components. It should be noted that parking structures with large plan
areasregardless of structural systemdid not perform as well as other
types of buildings.

Built according to good practices, concrete homes can be among the

safest and most durable types of structures during an earthquake. Homes
built with reinforced concrete walls have a record of surviving
earthquakes intact, structurally sound and largely unblemished. In
reinforced concrete construction, the combination of concrete and steel
provides the three most important properties for earthquake resistance:
stiffness, strength, and ductility.
Studies of earthquake damage consistently show
well-anchored shear walls are the key to
earthquake resistance in low-rise buildings.
Optimal design conditions include shear walls
that extend the entire height and are located on
all four sides of a building. Long walls are
stronger than short walls, and solid walls are
better than ones with a lot of openings for Properly anchored walls
windows and doors. These elements are are key to earthquake
designed to survive severe sideways (in-plane) resistance in low rise
forces, called racking and shear, without being buildings.
damaged or bent far out of position. Shear walls also must be well
anchored to the foundation structure to work effectively. Properly installed
steel reinforcing bars extend across the joint between the walls and the
foundation to provide secure anchorage to the foundation.

Low-rise buildings most vulnerable to earthquakes do not have the

necessary stiffness, strength, and ductility to resist the forces of an
earthquake or have walls that are not well anchored to a solid foundation,
or both. Three types of buildings sustain the most significant damage:
Multi-story buildings with a ground floor consisting only of columns;
Wood-frame houses with weak connections between the walls and
Unreinforced masonry or concrete buildings

Reinforced concrete walls work well because of the composite system:

Concrete resists compression forces, and reinforcing steel resists tensile
forces produced by an earthquake. Even a lightly reinforced concrete
shear wall has over six times the racking load resistance as framed wall

Blast resistance

Concrete has demonstrated blast resistance

through tests. The Insulating Concrete Form
Association (ICFA) and the Northern Virginia
Concrete Advisory Council successfully
demonstrated the blast-resistant properties of
ICF building systems during the Force Protection
Equipment Demonstration (FPED V) April 2628,
ICF reaction boxes prior to blast
2005, at Quantico Marine Corps Base in test
Northern Virginia. During the blast
demonstrations, eleven separate ICF reaction
boxes, weighing 13 tons apiece and with walls
measuring 8 feet tall and 6 inches thick were
subjected to explosion from 50 lbs of TNT at
differing distances (3.5 feet to 10 feet) and to
pressures from 300 pounds per square inch (psi)
to over 7,000 psi. Known for decades for its
impact resistant properties, expanded
An ICF wall after a 50 lb.
polystyrene (psi), the primary material in ICFs, TNT detonation from 10
has recently shown great potential as a blast- feet away.
resistant product. In each instance during six
different blast demonstrations, EPS compressed against the face of the
concrete wall and reduced the pressure of the blast.

In addition, high performance concrete can be designed to have improved

blast resistant properties. These concretes often have a compressive
strength exceeding 14,500 psi and contain steel fibers. These blast-
resistant structures are often used in bank vaults and military

Building Protection
Ubiquitous precast concrete planters provide protection to federal
buildings, museums, and national landmarks. These barriers are attractive
yet are a deterrent to wayward vehicles. Attractive concrete barriers that
also provide seating are becoming common.

A very large portion of digital image processing is devoted to image restoration. This
includes research in algorithm development and routine goal oriented image processing.
Image restoration is the removal or reduction of degradations that are incurred while the
image is being obtained. Visual information transmitted in the form of digital images is
becoming a major method of communication in the modern age, but the image obtained after
transmission is often corrupted with noise. The received image needs processing before it can
be used in applications. Image denoising involves the manipulation of the image data to
produce a visually high quality image. This thesis reviews the existing denoising algorithms,
such as filtering approach, wavelet based approach, and multifractal approach, and performs
their comparative study.

Reduction of noise is essential especially in the field of image processing. Several researchers
are continuously working in this direction and provide some good insights, but still there are
lot of scope in this field. Noise mixed with image is harmful for image processing. In this
dissertation we proposed an efficient Multithresholding approach for reducing noise and blur

In our approach we provide the comparison considering the image of Leena improve the SNR
ratio. Images are evermore corrupted with noise during acquisition, transmission, and
retrieval from storage media. Distinct dots in reality are stipple in a Photograph taken with a
digital camera under low lighting conditions. Abstract of sound is absolute especially in the
field of
Vision is the most advanced of human senses, so it is not surprising that images play the
single most important role in our perception. However, unlike humans, who are limited to the
visual band of the electromagnetic (EM) spectrum, imaging machines cover almost the entire
EM spectrum, ranging from gamma to radio waves. They can operate on images generated
by sources that human are not accustomed to associating with images. These include
ultrasound, electron microscopy and computer-generated images. Thus digital image
processing encompasses a wide and varied field of applications.

In this context several researches apply their work in this direction. Adaptive Directional
Lifting (ADL) is one of the image compressions due to the characteristics of representing the
edges and textures in images efficiently [1, 2]. Several researches have shown that the
application of image denoising can also benefit from this technique
[3, 4].

Because of this, it can effectively decorrelate the dependencies found over image
discontinuities and compact high frequency components induced by image features into the
lower level or low band pass. If we think about the Wavelet transform then it can be
effectively capture singular points up to two dimensions means including one dimension, but
it is fail in representing the major features like edge , color ,contour and so on. There are
several directional and non-directional redundant transforms which are explored in different
research papers, including the curvelet, contourlet, wedgelet, bandlet, and the steerable
wavelet [5-8].

1.1 Overview

There are several approaches which are basically concern with denoise an image data, such
as averaging filter, Median filter, Gaussian filter and Partial Differential Equations (PDE)
approach. If we analyze the properties of good images then it will be with the less noise and
minimize the blur or blur reduction is the important factor. The PDE approach is much
effective and applies in several research like [9],[10]. But it is more effective if we apply
fourth order partial differential equation. Applications of the PDE models can be widely
found in a broad range of image restoration tasks such as denoising and enhancement [11]
color image processing [12][13] and resolution. This provide us the future insight or work
with the forth order partial differential equation with the same order in the direction of blur

Image Denoising play an important role in Image processing task [14]. Remove the noise
when the edges are in the preserving state is called image denoising. In the image processing
task it is a major and most common problem. If we want a very high quality resolution
images as the outcome then we must consider the noise parameters for reducing those
parameters to achieve better. The main purpose or the aim of image denoising is to recover
the main image from the noisy image [15].

V(i)=U(i) + N(i)

Where v(i)is the observed value, u(i)is the true value and n(i) is the noise perturbation at a
pixel i. If we analyses then there are several ways of model the noise. In some of the cases the
noise is considerable. For modelling and framework purposes it is correct to additive white
Gaussian Noise (AWGN) which is adaptive in nature to model the noise parameters. For that
we also consider blur as the degrading performance categorization.
The Fast Fourier Transform (FFT) method is an analytical technique for numerical analysis. It
provide the approximate analysis to partial differential equations(PDE) which is generally
calculated by the expansion in terms of function , which is also called basis method and used
for calculating the unknown coefficients. Then we can apply the FFT method to partial
differential equations like
u 2u 2u
= c2( 2 + ) + E(x, y)
t x y 2
This reduces the number of spatial variables until only a two-point boundary-value problem
or initial-value problem remains, which is solved by standard methods. FFT is used for
separate the variables so that it is used in the elaboration of partial differential equation. The
PDE of Forth order is also help for reduction and separation because of the higher degree. It
is much more flexible in the context of uses and separations of variables, so the identification
is also easy and detectable and reduces the noise coefficients.
1.2 Objective and Motivation
The need for ecient image restoration methods has grown with the massive production of
digital images and movies of all kinds, often taken in poor conditions. No matter how good
cameras are, an image improvementis always desirable to extend their range of action.

A digital image is generally encoded as a matrix of grayscale or color values. Inthe case of a
movie, this matrix has three dimensions, the third one corresponding totime. Each pair (i,
u(i)), where u(i) is the value ati, is called a pixel, short for pictureelement. In the case of
grayscale images, i is a point on a two-dimensional (2D) gridand u(i) is a real value. In the
case of classical color images, u(i) is a triplet of valuesfor the red, green, and blue
components. All of what we shall say applies identicallyto movies, three-dimensional (3D)
images, and color or multispectral images.

The two main limitations in image accuracy are categorized as blur and noise.Blur is intrinsic
to image acquisition systems, as digital images have a nite number ofsamples and must
satisfy the ShannonNyquist sampling conditions. The secondmain image perturbation is

Each one of the pixel values u(i) is the result of a light intensity measurement,usually made
by a charge coupled device (CCD) matrix coupled with a light focusingsystem. Each captor
of the CCD is roughly a square in which the number of incomingphotons is being counted for
a xed period corresponding to the obturation time.When the light source is constant, the
number of photons received by each pixeluctuates around its average in accordance with the
central limit theorem. In otherwords, one can expect uctuations of order n for an incoming
photons. In addition,each captor, if not adequately cooled, receives heat photons. This
perturbation isusually called obscurity noise. In a rst rough approximation one can write
v(i) = u(i) +n(i),

where i I, v(i) is the observed value, u(i) would be the true value at pixel i,namely, the
one which would be observed by averaging the photon counting on a longperiod of time, and
n(i) is the noise perturbation. As indicated, the amount of noiseis signal-dependent; that is,
n(i) is larger when u(i) is larger. In noise models, thenormalized values of n(i) and n(j) at
dierent pixels are assumed to be independentrandom variables, and one talks about white
noise. So our main motivation of this dissertation to reduce the noise and blur of the image.

1.3 Problem Domain

Digital cameras produce three common types of noise: random noise, "fixed pattern" noise,
and banding noise. The three qualitative examples below show pronounced and isolating
cases for each type of noise against an ordinarily smooth grey background.Random noise is
characterized by intensity and color fluctuations above and below the actual image intensity.
There will always be some random noise at any exposure length and it is most influenced by
ISO speed. The pattern of random noise changes even if the exposure settings are identical.

Fixed pattern noise includes what are called "hot pixels," which are defined as such when a
pixel's intensity far surpasses that of the ambient random noise fluctuations. Fixed pattern
noise generally appears in very long exposures and is exacerbated by higher temperatures.
Fixed pattern noise is unique in that it will show almost the same distribution of hot pixels if
taken under the same conditions (temperature, length of exposure, ISO speed).

Banding noise is highly camera-dependent, and is noise which is introduced by the camera
when it reads data from the digital sensor. Banding noise is most visible at high ISO speeds
and in the shadows, or when an image has been excessively brightened. Banding noise can
also increase for certain white balances, depending on camera model.

Although fixed pattern noise appears more objectionable, it is usually easier to remove since
it is repeatable. A camera's internal electronics just has to know the pattern and it can subtract
this noise away to reveal the true image. Fixed pattern noise is much less of a problem than
random noise in the latest generation of digital cameras, however even the slightest amount
can be more distracting than random noise.

Knowledge about the problem domain is coded into the image processing system in the form
of knowledge database. This knowledge is as simple as describing the regions of the image
where the information of the interest is located. Each module will interact with the
knowledge base to decide about the appropriate technique for the right application. For
example, if the acquired image contains spike-like noise the preprocessing module interacts
with the knowledge base to select an appropriate smoothing filter-like median filter to
remove the noise.

After studying different approaches we observe that we can work with the reduction of noise
and SNR. There is also some scope for the reducing the time of denoising while unaffected
the accuracy. There are several approaches for noise free image retrieval but still there are lot
of scope in the direction of that better image will be retrieved. If we analyse the
research work in [14] and [15] they do not achieve good results in the case of noise
parameters , they do not detect any Blur + Noise combination when they consider the noise
parameters, it can be recovered by Thresholding Techniques for Image Denoising.. The noise
is also bit higher which can be reduces with blur and noise parameters with different noise
parameters and conditions.

1.4 Scope
After studying and analyzing several research works in the direction of image denoising, we
can suggest some following points which can be improved or there is the need of betterment
in the field of image denoising. The points are following:
1. Reduction of Noise with different noise parameters.
2. Need of reducing Blur
3. Reduction of time of denoising while unaffected the accuracy.
4. Betterment in SNR and PSNR.
5. Image reconstruction is also in less time with high quality accessibility.

1.5 Contribution of Thesis

Images are a natural way for humans to think about spatial information, and digital images
are a natural representation of spatial data. Like all recorded signals, digital images are often
corrupted by noise, increasing the difficulty with which human observers or computer
algorithms are able to extract the useful underlying information. Although noise can be
mitigated by improved image acquisition hardware, in some modalities, such as coherent
imaging, the noise is an inherent part of the imaging process.

However, information content may be preserved even at high noise levels. With some effort,
one can still discern the structural details in the presence of noise. Thus, the goal of noise
filtering, or image denoising, is to exploit the available information in the observed image to
obtain an estimate of the noise-free signal.

1.6 Structure of Thesis

This thesis documentation consists of various chapters which are given as follows:

Chapter 1 Introduction illustrates overview and objective of thesis. Then motivation and
problem definition are described. It then gives structure of the thesis document.

Chapter 2 Literature Survey provides the background necessary for the rest of the
dissertation. It reviews related concepts in the context of the work presented in this
dissertation. A brief survey about image denoising is presented. Main contribution of the
research presented in this dissertation is to establish connection for several applications; we
then focus our review on several novel technologies that are adopted by our method to
address interaction and infrastructure issues.

Chapter 3 Analysisillustrates overview and basis of the different denoising method with their
applicability. It provides the semantics of different services and uses.

Chapter 4 Proposed Approach discusses the software and the proposed work with proper
explanation with flowchart.In this chapter we also discuss the result analysis.

Chapter 5 Implementation Detail covers complete implementation detail of the application.

This chapter starts with overview of the implementation of the whole application. A result
illustrates the results obtained by using snapshots along with some explanations.

Chapter 6 Conclusion and Future work discusses about the conclusion drawn from
proposed dissertation work as well as some insights into future research on this topic.

2.2.3 Image Formats

There are various formats of the image supported by MATLAB for processing based on
requirements of application. The extension of the image file shows the format of the image.
The Table 2.1 describes various image formats.

Table 2.1: Image Formats supported by MATLAB

Format Name Description Extension

.tif, .tiff
TIFF Tagged Image File Format
.jpg, .jpeg
JPEG Joint Photograph Experts Group
GIF Graphics Interchange Format
BMP Window Bitmap
PNG Portable Network Graphics
XWD X Window Dump

2.3 Classes and Image Types

2.3.1 Data Classes
Although we work with integer coordinates, the values (intensities) of pixels are not restricted
to be integers in MATLAB. Table 1.1 lists the various classes supported by MATLAB and
the Image processing Toolbox for representing pixel values. The first eight entries in the table
are referred to as numeric data classes. The ninth entry is the char (character) class and, as
shown, the last entry is referred to as logical data class.
All numeric computations in MATLAB are done in double quantities, so this is also a
frequent data class encounter in image processing applications. Class uint8 and logical are
also encountered frequently, especially when reading images from image file formats such as
TIFF or JPEG. These classes use 1 byte to represent each pixel. Some scientific data sources,
such as medical imagery, require more dynamic range than is provided by uint8, so the
uint16 and int16 classes are used often for such data. These classes use 2 bytes for each array
element. The floating-point classes double and singleare used for computationally intensive
operations such as Fourier transform. Double-precision floating-point uses 8 bytes per
element, whereas, single-precision floating-point uses 4 bytes. The int8, uint32, and int32
classes, although supported by the toolbox, are not used commonly for image processing.
Many Image processing Toolbox functions however support all the data classes listed in
Table 2.2: Classes used for Image Processing in MATLAB

Name Description

Double Double precision, floating-point numbers in the approximate range 10308 (8 bytes
per element)

Single Single precision, floating-point numbers in the approximate range 1038 (4 bytes
per element)

uint8 Unsigned 8-bit integers in the range [0,255] (1 byte per element)

uint16 Unsigned 16-bit integers in the range [0,65535] (2 byte per element)

uint32 Unsigned 32-bit integers in the range [0,4294967295] (4 byte per element)

int8 Signed 8-bit integers in the range [-128, 127] (1 byte per element)

int16 Signed 16-bit integers in the range [-32768, 32767] (2 byte per element)

int32 Signed 32-bit integers in the range [-2147483648, 2147483647] (4 byte per element)

Char Characters (2 bytes per element)

logical Values are 0 or 1 (1 byte per element)

The char data class holds characters in Unicode representation. A character string is merely
a 1*n array of characters. Logical data class contains only the values 0 or 1, with each
element being stored in memory using function logical or by using relational operators.
MATLAB also supports uint64 and int64, but the toolbox does not.

2.4.2 Gaussian Noise

Gaussian noise is evenly distributed over the signal [Um98]. This means that each pixel in the
noisy image is the sum of the true pixel value and a random Gaussian distributed noise value.
As the name indicates, this type of noise has a Gaussian distribution, which has a bell shaped
probability distribution function given by,

1 ( X )2
f (X ) exp X

2 2 2

Where g oir X represents the gray level, m or is the mean or average of the function, and
is the standard deviation of the noise. Graphically, it is represented as shown in Figure 2.1.
When introduced into an image, Gaussian noise with zero mean and variance as 0.05 would
look as in Figure 2.4. Figure 2.5 illustrates the Gaussian noise with mean (variance) as 1.5
(10) over a base image with a constant pixel value of 100.

Figure 2.1Gaussian Distribution

Figure 2.2: Gaussian noise Figure 2.3: Gaussian noise

(mean=0, variance 0.05) (mean=1.5, variance 10)

2.4.3Salt and Pepper Noise

Salt and pepper noise [Um98] is an impulse type of noise, which is also referred to as
intensity spikes. This is caused generally due to errors in data transmission. It has only two
possible values, a and b. The probability of each is typically less than 0.1. The corrupted
pixels are set alternatively to the minimum or to the maximum value, giving the image a salt
and pepper like appearance. Unaffected pixels remain unchanged. For an 8-bit image, the
typical value for pepper noise is 0 and for salt noise 255. The salt and pepper noise is
generally caused by malfunctioning of pixel elements in the camera sensors, faulty memory
locations, or timing errors in the digitization process. The probability density function for this
type of noise is shown in Figure 2.6. Salt and pepper noise with a variance of 0.05 is shown
in Figure 2.7.

Figure 2.4PDF for salt and pepper noise

Figure 2.5Salt and pepper noise

2.4.4 Speckle Noise

Speckle noise is a multiplicative noise. This type of noise occurs in almost all coherent
imaging systems such as laser, acoustics and SAR(Synthetic Aperture Radar) imagery. The
source of this noise is attributed to random interference between the coherent returns. Fully
developed speckle noise has the characteristic of multiplicative noise. Speckle noise follows a
gamma distribution and is given as

where variance is
a and g is the gray level.
On an image, speckle noise (with variance 0.05) looks as shown in Figure 2.8.The gamma
distribution is given below in Figure 2.9.

Figure 2.6Gamma distribution

Figure 2.7Speckle noise

2.4.5 Brownian Noise

Brownian noise comes under the category of fractal or 1/f noises. The mathematical model
for 1/f noise is fractional Brownian motion. Fractal Brownian motion is a non-stationary
stochastic process that follows a normal distribution. Brownian noise is a special case of 1/f
noise. It is obtained by integrating white noise. It can be graphically represented as shown in
Figure 2.10. On an image, Brownian noise would look like Image 2.11 which is developed
from Fraclab.

Figure 2.8Brownian noise distributions

Figure 2.9Brownian noise

4.1 Combinational Design
The active research in image processing is noise. If we think about the corrupted images, then
we analyses that it is corrupted by random variations inintensity values which is the noise. It
is because of the data acquisition process. The main aim of image denoising methods is to
recover the original image or fetching the better quality image after reduction from a noisy
one, in order to perform, in an easier and with a more semantic way to a task which is the part
of image processing as image segmentation.

4.2 Structure
The same procedure employed for 1-D signal denoising can also be applied to image
denoising.After implementing the double-density DWT, real double-density dual-tree DWT,
and complex double-density dual-tree DWT for 2-D signals, we can develop three different
methods using these DWTs to remove noise from an image. The double-density DWT
method will be discussed first.

function y = double_S2D(x,T)
% x: noise signal
% T: threshold
[af, sf] = filters1;
J = 4;
w = double_f2D(x,J,af);
% loop thru scales
for j = 1:J
% loop thru subbands
for s = 1:8
w{j}{s} = soft(w{j}{s},T);
y = double_i2D(w,J,sf);
This program method takes two input parameters, the first being the noisy image, whose
dimension is 512 x 512, and the second being the threshold point. We take the forward DWT
over four scales and apply soft thresholding to the wavelet coefficients through all the
subbands. After thresholding, we then take the inverse wavelet transform.
Figure 4.1 Noisy Image
The following example shows how to convert an image to double data type, how to create a
noisy image and display the denoised image. Note that we use a threshold value of 20, which
is the optimal threshold point for this case.

Example (Noise Attenuation)

s1 = double(imread('peppers.jpg')); % load image as a double
s = s1(:,:,3); % convert to a 2-D image
figure(1) % display original image
axis image
title('Original Image')
x = s + 20*randn(size(s)); % add Gaussian noise to image
figure(2) % display noisy image
axis image
title('Noisy Image')
T = 15; % choose a threshold of 15
y = double_S2D(x,T); % denoise using Double-Density DWT
figure(3) % diplay denoised image
axis image
title('Denoised Image')
This program produces the following denoised image:

Figure 4.2 Denoised Image By Double-Density DWT Method.

From the resulting image, we can see the denoising capability of 2-D double-density DWT.
function y = doubleden_R2D(x,T);
[Faf, Fsf] = FSdoubledualfilt;
[af, sf] = doubledualfilt;
J = 4;
w = doubledualtree_f2D(x,J,Faf,af);
% loop thru scales:
for j = 1:J
% loop thru subbands
for s1 = 1:2
for s2 = 1:8
w{j}{s1}{s2} = soft(w{j}{s1}{s2},T);
y = doubledualtree_i2D(w,J,Fsf,sf);
This program results in the following denoised image:

Figure 4.3 Denoised Image By Double-Density Dual-Tree Real DWT Method

From the resulting image, we can see the denoising capability of 2-D double-density

function y = doubledual_C2D(x,T)
[Faf, Fsf] = FSdoubledualfilt;
[af, sf] = doubledualfilt;
I = sqrt(-1);
J = 4;
w = cplxdoubledual_f2D(x,J,Faf,af);
% loop thru scales
for j = 1:J
% loop thru subbands
for s1 = 1:2
for s2 = 1:8
C = w{j}{1}{s1}{s2} + I*w{j}{2}{s1}{s2};
C = soft(C,T);
w{j}{1}{s1}{s2} = real(C);
w{j}{2}{s1}{s2} = imag(C);
y = cplxdoubledual_i2D(w,J,Fsf,sf);

The program results in the following denoised image:

Figure 4.4 Denoised Image By Double-Density Dual-Tree Complex DWT Method

We can see that 2-D double-density method is best in terms of noise attenuation by the
following "RMS Error vs. Threshold Point" plot.
Figure 4.5 2-D Threshold Comparison
4.3.3 Flow Chart

As shown in the below flowchart. We proposed a multiple threshoding technique. First the
received noisy image is applied to first thresholding method and oupput of first thresholding
method is applied to second threshold in technique as shown in figure 4.6. In the next
subsequent section we also provide the result comparison which shows the effectiveness of
our approach.

Noisy Image


Bayes Threshold




Denoised Image

Figure 4.8 Working Process

Figure 4.8 explains the proposed method flow chart in detail. As shown in first stage discrete
wavelet transform of noisy image is taken and thresholded using Bayesian approach because
Bayesian threshold has the better performance in comparison to normal and visu threshold
technique. After applying threshold inverse discrete wavelet transform is taken. This is the
end of first level thresholding. This output acts as input for next stage of thresholding. In the
previous equation, M and N are the number of rows and columns in the input images,
respectively. Then the block computes the PSNR using the following equation:
PSNR = 10 log R 2 / MSE
In the above equation, R is the maximum fluctuation in the input image. For example, if the
input image has a double-precision floating-point data type, then R is 1. If it has an 8-bit
unsigned integer data type, R is 255, etc. Table- 1 compares various techniques on the basis
of their PSNR & MSE value for two types of noises salt & pepper noise and Gaussian noise.
The value of noise density for which values are calculated here is v=0.01. Figure 3 shows the
result for all the techniques. After the analysis of proposed scheme it is observed that it has
better performance for Gaussian noise in comparison to salt & pepper noise.

4.4 Result Evaluation

We used the Matlab R2010a to run the experiment under the PC environment of Windows 7
ultimate and CPU Pentium Dual-core which frequency is 1.86 GHz and memory with 2 GB.

For result comparison we consider different imagesof leena and show the effectiveness of
our algorithm. The proposed new denoising technique is compared with various existing
technique based on PSNR (Peak signal to noise ratio) and MES (Mean square error). The
Mean Square Error (MSE) and the Peak Signal to Noise Ratio (PSNR) are the two error
metrics used to compare image quality. The MSE represents the cumulative squared error
between the noisy and the original image, whereas PSNR represents a measure of the peak
error. The lower the value of MSE, lower the error.
To compute the PSNR, the block first calculates the mean-squared error using the equation:
MSE = (1/ MN) (m=1 to M) (n=1 to N) ( x(m,n)-x^ (m,n))2
The table 4.1 shows the comparison which shows the effectiveness of our approach.
Table 4.1: Result Analysis
Techniques Salt And Pepper Noise Gaussian Noise


BayesShrink 74.57 0.0022 71.65 0.0028

VisuShrink 74.37 0.0023 72.45 0.0057

Bilateral 73.53 0.0028 69.45 0.0029

Multiscale Thresholding 74.21 0.0024 73.10 0.0045

Proposed Method 75.14 0.0019 73.75 0.0027

In the above table when we compares various techniques on the basis of their PSNR &
MSE value for two types of noises salt & pepper noise and Gaussian noise. The value of
noise density for which values are calculated here is v=0.01. Figure 3 shows the result for all
the techniques. After the analysis of proposed scheme it is observed that it has better
performance for Gaussian noise in comparison to salt & pepper noise. The result is simulated
using Matlab.

6.1 Conclusion

In this dissertation we survey several aspects for image denoising. The process of removing
noise from an image is known as noise reduction or denoising. A standard denoising
technique is the convolutions of the image with different distribution technique. There is also
a need of we consider a new partial differential equations (PDE) denoising method which can
smooth out the high frequency oscillation while keeping the edges in the high noisy level

Then we proposed in this dissertation paper incorporates dual threshoding technique

which combines Bayes Shrinkage thresholding and Adaptive Multi-scale Products
Thresholding. This method multiplies the adjacent wavelet sub-bands to strengthen the
significant features in the image and then applies the thresholding to the multi scale products.
The advantage of proposed technique is that it has not only improvement in image
characteristics such as PSNR & MSE but also in image quality (Visual image). This
technique is also less complex in terms of level of denoising in comparison to adaptive
spatial multi-scale technique. The proposed technique is verified for Gaussian noise and salt
& pepper noise and observed that it has better performance for both but as the density of
noise is increased performance for salt & pepper noise is less but for Gaussian noise it has
significant performance.

6.2 Future Work

In future we can apply neural network and fuzzy sets to improve the results. We also apply
heterogeneous framework to handle the database which can make the database handling in
the proper way. We also try the approach with other databases and video files also.
[1] W. P. Ding and F. Wu, Adaptive directional lifting based wavelet transform for image
coding, IEEE Trans. Image Processing, vol. 16, no. 2, pp. 416684, 2007

[2] Y. Liu and K. N. Ngan, Weighted adaptive lifting-based wavelet transform,IEEE

Trans. Image Processing, vol. 17, no. 4, pp. 500511, 2008.

[3] X. T. Wang, G. M. Shi, and Y. Niu, Image denoising based on improved adaptive
directional lifting wavelet transform, in International Conference on Signal Processing,
2008, vol. 2, pp. 11121116.

[4] X. T. Wang, G. M. Shi, Y. Niu, and L. Zhang, Robust adaptive directional lifting
wavelet transform for image denoising, IET Image Process (Accepted), 2009.

[5] G. Y. Chen, B. Kegl, Image denoising with complex ridgelets,Pattern Recognition,

vol. 40, 2007,pp. 578-585,.

[6] Z. Liu, H. Xu, Image Denoising with Nonsubsampled Wavelet-Based Contourlet

Transform, Fifth International Conference on Fuzzy Systems and Knowledge Discovery,
2008, pp. 301-305.

[7] J. R. Sveinsson, Z. Semar, J. A. Benediktsson, Speckle Reduction of SAR Images in the

Bandlet Domain, IEEE International Geoscience and Remote Sensing Symposium, 2008,
pp. 1158-1161.

[8] B. Fisch and E.L. Schowart, "Iearning an Integral Equation Approximation to Nonlinear
Anisotropic Diffusion in Image processing", Dept. cognitive and Neural Systems Boston

[9] Perona. P and Ma[ik. J, "Sca[e-space and edge detection using anisotropic diffusion," in
Proceeding of IEEE Computer Society workshop on Computer Vision, 1987, pp. 16-27.

[10] Y. You, W. Xu, A. Tannenbaum and M. Kaveh, "Behaviora[ analysis of an isotropic

diffusion in image processing", IEEE Trans. Image Processing, vol. 5, 1996, pp. 1539-1553.