International Journal of Computer Applications (0975 – 8887)

Volume 177 – No. 38, February 2020

A Study on Image Compression and its Applications

B. Reddaiah
Department of Computer Applications
Yogi Vemana University
Kadapa, A.P, India

ABSTRACT Zip, Gzip, RAR, and Stufflt and 7z. An exclusive algorithm
The fast development in digital filed gave scope for a lot of for compression is used by each of the compression types for
image processing applications. To make use of digital images compressing the data.
efficiently, explicit methods are required to decrease the
number of bits essential for their presentation. This guided for
1.2 Media Compression
instant development in area of Image Processing. This As of file compression, the aim of compressing media file is
increase forced to develop more strong algorithms. to decrease the size of file so that the space in the disk can be
Compression concentrates on dropping image dimension saved. On the other hand, media compression algorithms are
without losing the originality and information present with explicit to particular media kinds, like images, audio, and
original one. With this goal of compressing the image is to video files.
reduce redundancy present in image and to increase the
storage capacity for well-organized communication. This 2. NEED OF IMAGE COMPRESSION
research reviews important image compression methods. This For single colored images there is no need of huge storage
paper discusses efficient techniques like lossy and lossless space to store the raw data. As the number of images to be
with its applications and research possibilities. stored increases or images to be transmitted increases the
entire storage requirements become irresistible. Along with
Keywords storage size, bandwidth for transmission and time taken for
Compression, Techniques, transmission, Lossy, Lossless, transmission of uncompressed images will be more and if
data, information, compressed before its storage and transmission it may be
proportional to the size of image. Before compressing the
1. INTRODUCTION images as an initial step different types of redundancies are to
Image or data compression is some form of compression that be eliminated. In common, three basic digital images
is used to decrease the size of any kind of file. When the redundancies are in existence and the redundancies are as
memory size of a file is reduced it requires less space to store follows [3].
and facilitates transmitting images in less time [2]. The
fundamental form of compression process is shown in 2.1 Spatial Redundancy
Fig 1. The values of neighboring pixels are strongly correlated in the
majority of the natural images. From the values of
neighboring pixels value of any given pixel can be predicted
practically. This prediction by using any individual pixel is
comparatively less significant and the redundancy can be
reduced. The capability of the wavelets transformation is to
efficiently capture variations at unusual scales. So, these
wavelets are preferably suitable for this purpose.

2.2 Coding Redundancy

An image that is stored in the form of matrix, then the values
range in between 0 to 255 and this frequency distribution of
pixel values is called histogram. With this storage
representation it is easy to find repeated times that each digit
happens in the image. In standard images, some pixel values
have larger frequency when compared to other pixels. In such
cases coding redundancy is incurred when the same code
word size is allocated to each pixel. Such redundancy can be
Fig: 1. Overview of Image Compression
reduced using Huffman coding. This is by assigning smaller
Data Compression methods are basically of two types. The number bits to the extra possible gray scale values than the
two methods are compression the data file and compression less possible bits.
the media file.
2.3 Psycho visual Redundancy
1.1 File Compression Psycho visual redundancy is based on the properties of human
In this type of compression the size of a file or size of more vision. Human eyes cannot respond to all visual information
number of files will be reduced. Once file compression is with equivalent sensitivity. Certain information is
made on a file or files the consequential collection may reduce comparatively less importance when compared to additional
the disk space from 50 percent to 90 percent than that of information and typical visual processing is called psycho
original form. Different types of file compressions include visual redundancy.

33
 International Journal of Computer Applications (0975 – 8887)
Volume 177 – No. 38, February 2020

3. COMPRESSION TECHNIQUES
Compression is a procedure where particular information is
characterized with reduced number of data points. Image
compression techniques as shown in Figure 2 and are broadly
classified into two categories as lossy and lossless image
compression [5, 4, 1].

Fig: 4. Overview of Lossless Compression

Lossless compression can not only used for graphic images
but it can be used for computer data files such as spreadsheets,
text documents and other software applications like sending
files as email attachments. While compressing text files for
repeated letters a common format is used and for the short
term they are discarded. Once the text file is decompressed,
Fig: 2. Types of Data Compression Methods the letters are retrieved.
These two different techniques have same aim and are used to The major advantages of Lossless Compression techniques
compress files. The Lossy and lossless techniques has are that they are mainly useful for compressing large size
different methods that are used by different file formats for files. Once the file is compressed when restoring the file back,
achieving results. Lossy compression methods are like exact data is restored after re-compression. After re-
Discreet Cosine Transform method, vector quantization compression high quality data is retrieved, so that there will
method and Huffman coding method whereas Lossless be no difference between the original image and restored
compression methods like Run Length Encoding method and image.
string table method. The compression is measured by the
compression ratio ( . The major disadvantages of Lossless Compression techniques
are that once the image is compressed the ratio of
Mathematically the quantity that measures the quality of a compression will be low. After compression when the data is
reconstructed image is compared with the original image and being transmitted the time taken to transfer the image is high.
is known as peak signal to noise ratio (PSNR), that is While reconstructing the image back after transmission
measured in decides (dB). decoding the image to get back the original image is a
difficult process.
3.1 Lossless Compression
This technique is generally applied for archival purposes. 3.1.1 Run Length Encoding
Images that have geometric shapes and are comparatively Run length encoding is a lossless compression technique that
simple are considered for lossless compression. It is used for replaces data by a pair of values namely value that indicates
images that belongs to medical, technical drawings, clip art the repeated value and length that indicates the number of
etc. In this compression technique the original data is repetitions. This compression technique is a simple one in
absolutely reconstructed from the compressed data and used which data is in the form of runs. Runs are sequences where
to size the file with no loss in image quality and is shown in same data value takes place in several successive data
the Figure 3. elements. They are stored as a single data value and count,
rather than as the original run.

3.1.2 Predictive Coding

This technique predicts the value of each pixel by using the
value of its neighboring pixels. As a result, each pixel is
encoded with a prediction error rather than its original value.
These errors to a great extent are smaller when compared to
original value, so that lesser bits are required to store bits.
Along with predictive coding another alternative predictive
coding is adaptive coding. This technique splits the image into
blocks. The prediction coefficient is computed independently.

3.1.3 Arithmetic Encoding

This encoding technique represents a message as some finite
intervals between 0 and 1. Based on the probabilities of the
Fig: 3. No loss shown in Lossy Compression message symbols the interval is divided into smaller intervals.
This technique is the most capable method to code symbols
While compressing all the information in the original images according to the probability of their occurrence.
is used. This will help in decompression to get back to
original form that is accurately equal to the input data and the 3.1.4 Entropy Encoding
process is shown in the Figure 4. This technique minimizes the size of dataset required to
communicate a particular amount of information. Entropy

34
 International Journal of Computer Applications (0975 – 8887)
Volume 177 – No. 38, February 2020

coding method is used in Huffman and arithmetic coding The files compressed by these methods are easily usable,
techniques. When there are repeated characters and symbols quick to open and good for using in internet applications. The
first they are noted down. After noting instead of repeating at processed file are with low resolution, low quality and mostly
every pixel, the positions of these pixels are recorded. Further free from threats.
same symbols are assigned.
The major disadvantages of Lossy Compression techniques
3.1.5 LZW Coding are that they are not supportive for editing the images. In these
LZ technique replaces repetitive substrings to previous images the editable data is dumped and have very minute
occurrences of strings. The repeated data is substituted by archive value. The resolution of the processed images will be
position and length of existing substring low by which the quality of printing the images may be
limited.
3.1.6 Huffman Coding
Huffman Coding technique makes use of variable length code. 3.2.1 Discrete Cosine Transform
Here short code words are allocated to most familiar values or Discrete Cosine Transform is a lossy Compression technique
for symbols in the data. Longer code words are allocated to specifically used for compressing image and audio files. This
less repeated occurring values. This technique generates less technique translates data into the summation of series of
redundancy cods when compared to other techniques. This cosine waves oscillating at different frequencies.
coding is used for compressing text, images, and videos.
3.2.2. Discrete Wavelet Transform
3.2 Lossy Compression Discrete wavelet transform characterizes image as a sum of
Lossy compression methods are particularly used to compress wavelet functions. These functions are known as wavelets that
natural images like photographs, where negligible loss for have dissimilar location and scale. This technique is typically
reliability is adequate to accomplish a considerable reduction executed by means of hierarchical filter structure and it is
in bit rate. This is most frequently used to compress applied on image blocks.
multimedia data and some data may be lost. This is not suited 3.2.3. Fractal Compression
and used for text based documents and software as they need
This technique depends on certain parts of images resembling
to keep the exact information without loss. This technique
other parts of same image. This algorithm converts these parts
allows higher compression ratio than the lossless technique. In
into mathematical data. This data is known as fractal codes
this compression the rebuilding of an image is only an
that are used to reconstruct the encoded image. As the image
approximation of the original data and original image is not
is converted into fractal code it becomes resolution
retrieved as it is and very small amount of data loss may be
independent.
seen as shown in Figure 5.
3.2.4. Vector Quantization
This technique is a fixed to fixed length algorithm called
LBG-VQ algorithm. It divides a large set of data points called
vectors into groups. It encodes values from a
multidimensional vector space into a finite set of values to
from a discrete subspace of lower dimension. This lower
space vector needs less storage space and by which data is
compressed. This technique is used for voice and images.

4. APPLICATIONS OF COMPRSSION
For image compression applications the above discussed
Fig: 5. Data loss shown in Lossy Compression Lossy Compression and Lossless compression methods are
used. Various techniques and algorithms are used in
This type of compression also looks for redundant pixel compressing images by different image compression
information and that is permanently discarded. So when the applications. Among these two methods, selection of a
file is retrieved back from compression the original data is not method to compress images depends on the quality of the
exactly retrieved. Even though when the exact data is not output that is expected by the users. When a very high quality
retrieved, data missing cannot be experienced. The process of output is expected without any loss of data by users from
lossy compression is shown in Figure 6. image compression applications then lossless compression
technique can be used by the application.
Lossless compression techniques are used when a high degree
of accuracy is needed. In some cases where the quality is not
that much important and can be compromised, lossy
compression technique can be used. When this compression
technique is used there will be very little bit of loss in quality
and that loss cannot be visible to the human eye at most of
times. It can be used in such applications where a little
compromise on quality of image is tolerable. The areas where
image compression is used are in television broadcasting,
Remote sensing via satellite, for military communication
Fig:6 Overview of Lossy Compression systems through radars, Tele conferencing systems,
The major advantages of Lossy Compression techniques are communications systems built through computers, Facsimile
that they are mainly useful for compressing small files. Once transmission medical images in computer tomography
the files are compressed, they take very less time to send the magnetic resonance imaging, capturing and transmitting
files from one place to another and easy to move and store. satellite images, geological surveys, weather reporting

35
 International Journal of Computer Applications (0975 – 8887)
Volume 177 – No. 38, February 2020

applications. Software Engineering, Volume 4, Issue 7, July 2014.

5. CONCLUSION [2] Harpreet Kaur, Rupinder Kaur, Nvdeep Kumar, “Review

In this study different image compression methods has been of Various Techniques for Medical Image Compression”,
discussed. There is wide research going on in the field of International Journal of Computer Applications, Volume
compression. This is possible by keeping in view the 123, No. 4, Auguset 2015.
requirement of low bit rate methods for compression this [3] Bhonde Nilesh, Shinde Sachin, Nagmode Pradip, D.B.
study has been carried out. This study has found that there is Rane, “Image Compression Using Discrete Wavelet
more scope for research in this filed to develop more efficient Transform”, IJCTEE, Volume 3, March-April 2013.
algorithms.
[4] A. Alarabeyyat, S. Al-Hashemi, T. Khdour, M. Hjouj
There is more demand for compression algorithms and it is Btoush, S. Bani-Ahmed, R. Al-Hashemi, “Lossless
found that the research is to be made keeping in mind peak Image Compression Technique Using Combination
signal to noise ratio and compression ratio for better Methods”, Journal of Software Engineering and
algorithms. Applications, 2012.
6. REFERENCES [5] Malwinder Kaur, Navdeep Kaur, “A literature Survey on
[1] Richa Goyal, Jasmeen Jaura, “A Review of Various Losllless Image Compression”, Internatiional Journal of
Image Compression Techniques”, International Journal Advanced Research in Computer and Communication
of Advanced Research in Computer Science and Engineering, Vol., 4, Issue , March 2015.

IJCATM : www.ijcaonline.org 36