2017 Fourth International Conference on Image Information Processing (ICIIP)

Competency Assessment of Image Compression in

the Lossy and Lossless Domain
Devee Darshani Panda Debabala Swain
School of Computer Engineering School of Computer Engineering
KIIT University,Bhubaneswar KIIT University, Bhubaneswar
deveedarshani288@gmail.com debabala.swain@gmail.com

Abstract— Due to the data transmission and storage constraints

it is essential to compress the images before transmission. II. TECHNIQUES OF IMAGE COMPRESSION
Whereas the image quality gets degraded due to the reduced size The various type of image compression techniques are as
and information loss. So the recovered images endure with the follows:
corrupted image quality and high compression ratio. This paper
presents a strategic study on different image compression 1. Lossy Compression in images
techniques with their qualitative and quantitative analysis.The
2. Lossless Compression in images
performance evaluation of different lossy and lossless
compression techniques are done through a test image with
different performance measuring parameters like PSNR , MSE, A. Lossy Compression in images
CR and BPP. In this technique, once the image is compressed then the
recreated image will not be precisely identical to the original
image. Due to information loss many details of image data are
Keywords—Image compression; lossy compression; Lossless abandoned during compression. But appearance wise the
compression; Fractal Transformation; DCT; DWT; DPCM; image does not change in the exterior format As many pixel
Huffman; SPIHT; PSNR; MSE; CR; BPP. values get altered due to the applied technique for image size
reduction, hence it is called as lossy compression. Although
there is a remarkable change in the image quality using this
I. INTRODUCTION technique, still it is more useful and more applicable during
Visual information plays essential role in human life for data transmission. Many such techniques are explained in this
daily world communications. An image is represented in the section. Some of them are Discrete Wavelet Transform,
form of a two dimensional matrix of pixels. An image matrix Discrete Cosine Transform, Fractal transform, Transform
format is different for color and gray scale. As the image size coding etc [2].
is getting increased, the subsequent matrix also gets larger,
which in turn takes larger memory space for storage,
processing and more time consuming during transmission.
The gray images are represented in the form of matrices,
whose values range between 0 to 255, using 256 different
color codes. Among them the clean white and clean black
holds the color code 255 and 0 respectively. Due to the
limitations of transmission media like, bandwidth,
transmission speed, transmission time, storage space, it is
crucial to compress the image without degrading its quality of
content. Then it can be used in different applications. Further
the image can be again extracted after decompressing. Based
on the quality of the image, the compression technique can be
of two types, i.e., lossy and lossless compression [1].
This paper is organized as follows : Section II represents Fig. 1. Lossy image compression
the techniques of image compression in which lossy and
lossless compression are discussed, section III represents the
performance analysis of different techniques and finally a. Fractal Transform
section IV winds up the paper. Amongst the different image compression techniques,
fractal is a foremost asymmetric lossy technique. In this
method the original image is partitioned into different
segments using various methods like, edging, spectrum
analysis, color partitioning etc. There after each image

978-1-5090-6734-3/17/$31.00 ©2017 IEEE 508

 2017 Fourth International Conference on Image Information Processing (ICIIP)

segment is searched in the fractal library. During the process of c. DCT (Discrete Cosine Transform)
encoding and decoding , the fractal encoder is mainly used to Ahmed et al have first proposed DCT for image
convert a bitmap image to fractal codes. After the fractal compression. It is the most trendy method for image
conversion the bitmap image is able to change to fractal image transformation to compressed form. It is used for JPEG format
which can be divided into segments. Another gain is that for in lossy image compression. This compression method consists
any given input image of any size, the fractal codes can be of four stages of coding. In the first stage the original image is
stored in a smaller sized coded form than the original bigger partitioned into blocks of order zxz. In the second stage the sub
image. Afterwards the matching of the encoded fractal is dine block is transformed to frequency domain using 2D DCT
with the fractal code library. During matching the code function. With the process of quantization the frequency
matcher looks for the best fit match rather than the exact match coefficients are converted to lossless entropy coder. Due to the
based on different compression parameters. The performance proficient image compression, the original gets compact and
analysis in section-III shows that fractal compression is better the transformed energy parameter gets minimized [4]. The
than the other compression techniques in terms of compression wavelet transform can perform better than the DCT transform
ratio, BPP, MSE and PSNR [3]. with better compression ratio, which is discussed in the
B.Bhuvaneswari et. al have proposed a fast fractal image performance analysis section .
compression based on image interfacing system. Mohammed A.H.M. Jaffar Iqbal Barbhuiya et. al have proposed an
Ismail.B et. al have proposed an improved image compression image quality prediction for DCT based compressed. Ankit
technique using fractal transfrom for Color images by dividing Chouhan et. al have proposed an double compression of joint
the image into blocks with different size. photographic experts group(JPEG) image using discrete cosine
transform(DCT) technique with different quality factor(QF)
b. DWT (Discrete Wavelet Transform) for each step.
Another key method of image compression is discrete
wavelet transform which can be used as both lossy and B. Lossless Compression in images
lossless. The lossless DWT is used for qualitative compression Using Lossless Compression in images, the recreated
with better compression ratio. The DWT method transforms image will be identical to the original image as quantization
the discrete time signal to Discrete Wavelet Representation. never occurs. Most of the image information before
The DWT transformed data is based on time-scale compression are preserved after decompression, so it is known
representation, which can be further represented in multi- as 'lossless'[5]. This technique is most useful in many
resolution format. Wavelets transform is considered to be the applications, Run-length coding, DPCM, SPIHT etc.
most beneficial for image processing applications, which is
mainly used to suppress the redundant noise with blurring
effect. Wavelet transform has been the used for both image
and video compression due to its multi resolution image
analysis. Also it performs better due to noise removal, edge
detection, compression analysis of the image. In wavelet
transform an image of NXM can be decomposed into a part
with lower frequency estimation and other with higher
frequency estimation. The low pass filter uses the scaling
function and the high pass filter uses the wavelet function [3].

Fig. 3. Lossless Image compression

a. DPCM
Differential pulse-code modulation is a lossless
compression technique where the signal encoder uses the
pulse-code modulation for data coding also adds some more
utility of prediction based on pulse coded samples. The signal
Fig. 2. Wavelet Decomposition can be input in analog or digital format[6].

Afshan Mulla et. al have proposed an image compression

technique known as Cognitive Sub-Band Replacement
Discrete Wavelet Transform. M. Shaaban Ibraheem et. al have
proposed an medical image comprssion technique using
logarithmic DWT .

509
 2017 Fourth International Conference on Image Information Processing (ICIIP)

LSP (List of Significant Pixels) – Here the list of pixels are

found to have magnitudes larger than the threshold
(significant). The sorting pass is performed on the above three
lists [8].

III. PERFORMANCE ANALYSIS OF DISCUSSED

TECHNIQUES
Fig. 4. DPCM block diagram The performance analysis is begin discussed for different
lossy and lossless compression techniques.The performance
Rime Raj Singh Tomar et. al have proposed an method study is begin done through different parameters like PSNR ,
called Enhanced DPCM Transformation (EDT). Chen Zaho et. MSE, CR and BPP by taking the standard color lena image of
al have proposed an compressive sensed image coding via size 512X512.
stripe based DPCM.
TABLE I. LOSSY TECHNIQUES
b. Huffman Coding
Huffman coding represents another lossless image Parameters
PSNR MSE CR BPP
compression paradigm. It uses the Huffman source coding Methods
algorithm to generate the uniquely decipherable Huffman code DCT 85.90 1.67 17.91 0.12
with a minimum expected codeword length, when the DWT 52.91 0.03 6.92 1.15
probability distribution of a data source is known to the
encoder. Entropy represents the amount of information must Fractal 62.79 0.23 2.76 1.58
be coded by a compression algorithm. Low entropy images
contain lot of black sky with very little contrast. TABLE II. LOSSLESS TECHNIQUES
Mathematically it represents the amount of bits used in the
Parameters
image for coding during compression. Huffman coding uses a PSNR MSE CR BPP
specific method for choosing the representation for particular Methods
images which results in a prefix code. The Compression of DPCM 48.07 4.09 0.25 0.75
Images and data are both possible using Huffman Coding [3]. Huffman 26.13 2.85 27.89 1.58
Takiyah K. Cooper et. al have proposed an analysis of SPHIT 30.99 51.70 2.58 0.62
XOR filtered images method using huffman coding. Rachit
Patel et. al have proposed a fast and improved image
compression technique using huffman coding.

c. SPIHT Algorithm
Shapiro proposed the Set Partitioning In Hirecherical Trees
techniques in image compression . It uses the embedded zero
wavelet(EZW) algorithm for its functionalists. In this
algorithm the image undergoes the wavelet transformation.
Then it is decomposed into significant and insignificant
partitions based on the following function:

­1, max( i, j )  T ^ci, j `t 2n

SnT  ®
¯0, Otherwise
Here Sn(T) is the significance of a set of coordinates
T, and ci,j is the coefficient value at coordinate (i, j). Fig. 5. Original Image
There are two passes in the algorithm- the sorting pass and
the refinement pass [7]. The SPIHT encoding process uses
three lists.
LIP (List of Insignificant Pixels) –. Here the magnitudes
are smaller than the thresholds so it contains individual FRACTAL DWT DCT
coefficients.
LIS (List of Insignificant Sets) – The set of wavelet
coefficients are defined by tree structures and they are found to Fig. 6. Lossy Compressed Images
have magnitudes smaller than the threshold.

510
 2017 Fourth International Conference on Image Information Processing (ICIIP)

REFERENCES
[1] Rasha Adel Ibrahim, Sherin M. Youssef and Saleh Mesbah
Elkaffas,”An enhanced fractal image compression integrating quantized
quadtrees and entropy coding ” 11th International Conference on
Innovations in Information Technology , 2015 .
[2] Rime Raj Singh Tomar and Kapil Jain, “Lossless Image Compression
Using Differential Pulse Code Modulation and Its Application
”International Conference on Computational Intelligence and
Communication Networks, 2015.
DPCM Huffman SPIHT [3] Praisline Jasmi, R., B. Perumal, and M. Pallikonda Rajasekaran.
"Comparison of image compression techniques using huffman coding,
DWT and fractal algorithm",International Conference on computer
Communication and Informatics (ICCCI), 2015.
Fig. 7. Lossless Compressed Image
[4] Barbhuiya, A.H.M. Jaffar Iqbal, Tahera Akhtar Laskar, and K.
Hemachandran. "An Approach for Color Image Compression of JPEG
and PNG Images Using DCT and DWT", 2014 International
IV. CONCLUSION Conference on Computational Intelligence and Communication
This paper summarizes the different techniques on lossy Networks, 2014.
and lossless compression as follows. Using lossless [5] Ankit Chouhan and M.J. Nigam. "Double compression of JPEG image
using DCT with estimated quality factor",IEEE 1st International
compression each of the single pixel present in the original Conference on Power Electronics, Intelligent Control and Energy
image remains unchanged after uncompression so all Systems (ICPEICES), 2016.
information bits are completely restored. On the other hand a [6] Akshay Kekre and Dr. Sanjay Pokle, “Image Compression Using
lossy compression technique reduces an image size Wavelet Transform And Differential Pulse Code Modulation Technique
permanently by removing certain redundant information so ”,International Journal of Engineering Research & Technology (IJERT)
after uncompression many parts of the original image are lost Vol. 2 Issue 7, July - 2013
but it cannot be noticed by the user. This compression [7] Bhagya Raju, V., K. Jaya Sankar, and C. D. Naidu. "Real time multi
technique is used for those data where a certain information spectral image compression for high resolution imageries through super
resolution algorithm with DWTISPIHT",International Conference on
loss dose not affect to the user. The portion of information loss Electrical Electronics Signals Communication and Optimization
in lossy technique is decided by the user. However the efficient (EESCO), 2015.
compression technique on the user input image is to be chosen [8] Swetha Dodla, Y David Solmon Raju and K V Murali Mohan,”Image
by the user itself based on the type of image application. Compression using Wavelet and SPIHT Encoding Scheme” IJETT
,2013.

511