Image Compression and Reconstruction Using Artificial Neural Network

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 03 Issue: 02 | Feb-2016 www.irjet.net p-ISSN: 2395-0072
© 2016, IRJET ISO 9001:2008 Certified Journal Page 1617
Image Compression and Reconstruction Using
Artificial Neural Network
Ms. C. V. Yadav1, Ms. S. R.Gaikwad2, Ms. N. N. Jakhalekar3, Mr. V. A. Patil4
1B.E. Student, Dept. of E&TC Engineering, ADCET Ashta, Maharashtra, India
4Assistant Professor, Dept. of E&TC Engineering, ADCET Ashta, Maharashtra, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract: In this paper a neural network based image
compression method is presented. Neuralnetworksoffer
the potential for providing a novel solution to the
problem of data compressionbyitsabilitytogeneratean
internal data representation. This network, which is an
application of back propagation network, accepts a
large amount of image data, compressesitforstorageor
transmission, and subsequently restores itwhendesired.
A new approach for reducing training time by
reconstructing representative vectors has also been
proposed. Performance of the network has been
evaluated using some standard real world images. It is
shown that the development architecture and training
algorithm provide high compression ratio and low
distortion while maintaining the ability to generalize
and is very robust as well.
Key Words: Artificial Neural Network (ANN), Image
Processing, Multilayer Perception (MLP) and Radial
Basis Functions (RBF), Normalization, Levenberg-
Marquardt, Jacobian.
1. INTRODUCTION
Artificial Neural networks are simplified models of
the biological neuron system andthereforehavedrawntheir
motivation from the computing performed by a human
brain. A neural network, in general, is a highly
interconnected network of a large number of processing
elements called neurons in an architecture inspired by the
brain. Artificial neural networks are massively parallel
adaptive networks of simple nonlinear computing elements
called neurons which are intended to abstract and model
some of the functionality of the human nervous system in an
attempt to partially capture some of its computational
strengths. A neural network can be viewed as comprising
eight components which are neurons,activationstatevector,
signal function, pattern of connectivity, activity aggregation
rule, activation rule, learning rule and environment.
Recently, artificial neural networks [1] are
increasing being examined and considered as possible
solutions to problems and for application in many fields
where high computation rates are required[2].ManyPeople
have proposed several kinds of image compressionmethods
[3]. Using artificial neural network (ANN) technique with
various ways [4, 5, 6, 7]. A detail survey of about how ANN
can be applied for compression purpose is reported in [8, 9].
Broadly, two different categories for improving the
compression methods and performance have been

suggested. Firstly, develop the existence method of
compression by use of ANN technologysothatimprovement
in the design of existing method can be achieved. Secondly,
apply neural network to develop the compression scheme
itself, so that new methods can be developed and further
research and possibilities can be explored for future. The
typical image compression methods are based on BPNN
techniques. TheBack propagationNeural Network (BPNN)is
the most widely used multi layer feed forward ANN. The
BPNN consists of three or more fully interconnected layers
of neurons. The BP training can be applied to any multilayer
NN that uses differentiable activation function and
supervised training.
The BPNN has the simplest architectureofANN that
has been developed for image compressionbutitsdrawback
is very slow convergence. Mapping the gray levels of the
image pixels and their neighbors in such a way that the
difference in gray levels of the neighbors with the pixel is
minimized and then the CR and network convergencecan be
improved. They achieved this by estimating a Cumulative
Distribution Function (CDF) for the image. They usedCDFto
map the image pixels, then, the BPNN yields high CR and
converges quickly. In BPNN for image compression and
developed algorithm based on improved BP. The blocks of
original image are classified into three classes: background
blocks, object blocks and edge blocks, considering the
features of intensity change and visual discrimination
Finally, an adaptive method based on BPNN for image
compression/decompression based on complexity level of
the image by dividing image into blocks, computing the
complexity of each block and then selecting one network for
each block according toitscomplexityvalue.Theyusedthree
complexity measure methods such as: entropy, activity and
pattern-based to determine the level of complexity in image
blocks.
This paper is organized as follows. In section II we
discuss Methodology (Image compression using ANN) III
Describes the Neural network models. IV Describes the
multi-layer perception neural network anditsapproachthat
is directly developed for image compression. In section V
describe the Process steps for compression. VI explains the
experimental results of our implementation are discussed
and finally in section VII we conclude this research and give
a summary on it.
1.2 Image Processing:
Image Processing is a veryinterestinganda hotarea
where day-to-day improvementisquiteinexplicableand has
become an integral part of own lives. Imageprocessingis the
analysis, manipulation, storage, and display of graphical
images. An image is digitized to convert it to a form which
can be stored in a computer's memory or on some form of
storage media such as a hard disk. This digitization
procedure can be done by a scanner, or by a video camera
connected to a frame grabber board in a computer. Once the
image has been digitized, it can be operated upon by various
image processing operations. Image processing is a module
that is primarily used to enhance the qualityandappearance
of black and white images. It also enhances the quality of the
scanned or faxed document, by performing operations that
remove imperfections. Image processing operations can be
roughly divided into three major categories, Image
Enhancement, Image Restoration and Image Compression.
2. Artificial Neural Network (ANN):
The Soft Computing book by S. N. Shivanandam
gives the detail information about ANN. Artificial neural
networks are massively parallel adaptivenetworksofsimple
nonlinear computing elements called neurons which are
intended to abstract and model some of the functionality of
the human nervous system in an attempttopartiallycapture
some of its computational strengths. A neural network can
be viewed as comprising eight components which are
neurons, activation state vector, signal function, pattern of
connectivity, activity aggregation rule, activation rule,
learning rule and environment.

ANN has the special functionalitieslike,adaptability,
self learning capability. The ANN requires inputs with real
type and the sigmoid function of each ANN neuron requires
the input data to be in the range [0-1].
2.1 Training the ANN:
The input image is split up into blocks or
vectors of 4×4, 8 ×8 or 16×16 pixels. These vectors are
used as inputs to the network. The network is provide
by the expected (or the desired) output, and it is
trained so that the coupling weights, {wij}, scale the
input vector of N -dimensionintoanarrowchannelofY
-dimension (Y < N) at the hidden layerandproducethe
optimum output value which makes the quadratic
error betweenoutputandthedesiredoneminimum.In
fact this part represents the learning phase, where the
network will learn how to perform the task. In this
process of leering a training algorithm is used to
update network weights by comparing the result that
was obtained and the results that was expected.Itthen
uses this information to systematically modify the
weight throughout the network till it finds the
optimum weights matrix.
Fig.-1 Block diagram of ANN
3. Image Compression:
The information about Image Compression is
referred from IEEE paper on “Image Compression and
Reconstruction Using Artificial Neural Network” published
by K. Siva Nagi Reddy, Dr. B. R.Vikram, L. Koteswara Rao, B.
Sudheer Reddy.
Image compression techniques aim to remove the
redundancy present in data in a way, which makes image
reconstruction possible. Image compression continuestobe
an important subject in many areas such as communication,
data storage, computation etc.
In order to achieve useful compression various algorithms
were developed in past. A compression algorithm has a
corresponding decompression algorithm that, given the
compressed file, reproduces the original file. There have
been many types of compression algorithms developed.
These algorithms fall into two broad types, 1) Loss less
algorithms, and 2) Lossy algorithms. A lossless algorithm
reproduces the original exactly. Whereas, a lossy algorithm,
as its name implies, loses some data. Data loss may be
unacceptable in many applications.
Chart-1 Image Compression

4. Image Decompression:
To decompress the image; first the compressed
image is renormalized then applies it to the output of the
hidden layer and get the one vector of the hidden layer
output is normalized then it rasterization to represent the
reconstruct the image.
Chart-2 Image Decompression
5. CONCLUSIONS:
In this project the use of Multi -Layer Perception
Neural Networks for image compression is reviewed. Since
acceptable result is not resulted by compression with one
network, a new approach is used by changing the Training
algorithm of the network with modified LM Method. The
proposed technique is used for image compression. The
algorithm is tested on varieties of benchmark images.
Simulation results for standard test images with different
sizes are presented. These results are compared with L-M
method. Several performance measures are used to test the
reconstructed image quality.
According to the experimental results,theproposed
technique with modified L-M method outperformed the
existing method. It can beinferredfromexperimental results
as shown in Table 1, 2 and 3 that the proposed method
performed well and results higher compression ratio.
Besides higher compression ratio it also preserves the
quality of the image. It can be concluded that the integration
of classical with soft computing based image compression
enables a new way for achieving higher compression ratio.
REFERENCES:
[1] R. P. Lippmann, “An introduction to computing with
neural network”, IEEE ASSP mag., pp. 36-54, 1987.
[2] M.M. Polycarpou, P. A. Ioannou, “Learning and
Convergence Analysis of Neural Type StructuredNetworks”,
IEEE Transactions on Neural Network, Vol 2, Jan 1992,
pp.39-50.
[3] K. R Rao, P. Yip, Discrete Cosine Transform Algorithms,
Advantages, Applications, Academic Press, 1990
[4] Rao, P.V. Madhusudana, S.Nachiketh,S.S.Keerthi, K.
“image compression using artificial neural network”.EEE,
ICMLC 2010, PP: 121-124.
[5] Dutta, D.P.; Choudhury, S.D.; Hussain, M.A.; Majumder, S.;
”Digital image compression using neural network” .IEEE,
international Conference on Advances in Computing,
Control, Telecommunication Technologies, 2009. ACT '09.
[6] N.M.Rahim, T.Yahagi, “Image Compression by new sub-
image bloc ClassificationtechniquesusingNeural Networks”,
IEICE Trans. On Fundamentals, Vol. E83-A, No.10, pp 2040-
2043, 2000.
[7] M. S. Rahim, "Image compression by new sub- image
block Classification techniques using neural network. IEICE
Trans. On Fundamentals of Electronics, Communications,
and Computer Sciences, E83-A (10), (2000), pp. 2040-2043.
K.Siva Nagi Reddy, Dr.B.R.Vikram, L.Koteswara Rao,
B.Sudheer Reddy International Journal of Image Processing
(IJIP), Volume (6) : Issue (2) : 2012 85

BIOGRAPHIES
Ms. Charushila V. Yadav.
She is studying in Annasaheb
Dange College Of Engineering
and Technology, Ashta, MH,
India.
She is student of electronics
and telecommunication
department.
Ms. Shalaka R. Gaikwad.
India.
department.
Ms. Nilam N. Jakhalekar
India.
department.
Mr. Vikas A. Patil
He is working as Assistant
Professor in Annasaheb Dange
College Of Engineering and
Technology, Ashta, MH, India.
He is having 2 years and 7
months of teaching experience.
Area of specialization: Digital
System.

Image Compression and Reconstruction Using Artificial Neural Network

More Related Content

What's hot (18)

Similar to Image Compression and Reconstruction Using Artificial Neural Network (20)

More from IRJET Journal (20)

Recently uploaded (20)

Image Compression and Reconstruction Using Artificial Neural Network