The International Journal of Multimedia & Its Applications (IJMA) Vol.7, No.

2, April 2015

A COMPARATIVE STUDY ON ADAPTIVE

LIFTING BASED SCHEME AND INTERACTIVE
ARTIFICIAL BEE COLONY ALGORITHM FOR
IMAGE COMPRESSION
G.G Rajput1 and Vrinda Shivashetty2
1

Department of Computer Science, Rani Channamma University, Belagavi, India

2
Department of Computer Science, Gulbarga University Gulbarga, India

ABSTRACT
Adaptive update lifting scheme based Interactive artificial bee colony algorithm is proposed in this paper.
Wavelet transform based compression technique is used for images and multimedia files. Approximation
and detail coefficients are extracted from the signal by filtering in wavelet transform. To increase
frequency resolution both approximation and detail coefficients are re-decomposed up to some level.
Artificial bee colony algorithm by local search finds different update coefficients to get quality of
compressed image by choosing optimally best update coefficient. In IABC, the affection between employed
bees and the onlooker bees is found by considering the concept of universal gravitation. By passing on
control parameter different values, the universal gravitation involved in the IABC has a single onlooker
bee & variety of quantities of employed bees. As a result, IABC compared with existing image compression
schemes such as wavelet transform and Artificial Bee colony Algorithm, the proposed work gives better
PSNR.

KEYWORDS
Image Compression, Artificial bee colony, Wavelet Transform, Adaptive Lifting Scheme.

1. INTRODUCTION
Wavelet coding is recognised as efficient coding technique for lossy image compression. It is
well known that, the wavelet transform decomposes a typical image data to a few coefficients
with large magnitude and many coefficients with small magnitude. As most of the energy of the
image concentrates on these coefficients with large magnitude, lossy compression systems just by
using coefficients with large magnitude can realize the reconstructed image with good quality and
high compression ratio. For wavelet transforms, Lifting Scheme(LS) (3,4)allows efficient
construction of the filter banks. The restriction of this structure is that the filter structure is fixed
over the entire signal. In many applications to shape itself to the signal it is very much desirable
to design the filter banks. A number of such adaptive Lifting Schemes are proposed in the
literature which consider local characteristics of the signal for adapting [13,16]. Ramanathan et
al.[15] applied ABC algorithm to determine the optimal window size to obtain satisfactory
compression and quality in the multi-objective manner. M. Mohamed Ismail and Dr. K.Baskaran
DOI : 10.5121/ijma.2015.7205

57

The International Journal of Multimedia & Its Applications (IJMA) Vol.7, No.2, April 2015

[16] have implemented ABC algorithm in update process of lifting scheme to get better PSNR.
PEI-WEI TSAI et al[23] have proposed the Interactive Artificial Bee Colony Optimization
algorithm.
Image compression technique using adaptive update lifting scheme and Interactive Artificial Bee
Colony Algorithm (IABC) is presented in this paper. The results obtained are compared with
wavelet transform & Artificial Bee Colony Algorithm proposed in the literature. The rest of the
paper is described as follows. Section II provides a brief overview of basic image compression
techniques. In section III, a general lifting scheme is discussed and compared with the adaptive
lifting scheme. Section IV discusses about the wavelet transform. In Section V, Artificial Bee
Colony Algorithm is explained and section VI describes the Interactive Artificial Bee Colony
algorithm and Experimental results are discussed in Section VII.

2. BASIC IMAGE COMPRESSION TECHNIQUES

The image compression techniques are generally classified into two categories depending
whether or not an exact replica of the original image could be reconstructed using the compressed
image. These are:
1. Lossy Image Compression technique
2. Lossless Image Compression technique
1. Lossy Image Compression Techniques
Lossy schemes provide much higher compression ratios than lossless schemes. Lossy schemes
are widely used since the quality of the reconstructed images is adequate for most applications.
By this scheme, the decompressed image is not identical to the original image, but reasonably
close to it. The most popular current lossy image compression methods use a transform-based
scheme[25].
2. Lossless Image Compression Techniques
In lossless compression techniques, the original image can be perfectly recovered from the
compressed image. These are also called noiseless since they do not add noise to the signal. It is
also known as entropy coding since it use decomposition techniques to minimize redundancy[21].

3. LIFTING SCHEME
Lifting scheme is simplest and efficient algorithm to calculate wavelet transforms. Lifting scheme
is used to implement critically sampled filter banks which have integer output. The lifting scheme
can custom design the filters, essential in the transform algorithms. Independent of translating
and dilating, needless of frequency analysis, lifting scheme is processed into space domain. An
answer to the algebraic stage of wavelet construction is provided by lifting scheme, which leads
to a fast in-place calculation of the wavelet transform, i.e. it does not require auxiliary memory.
Different wavelets show different image compression effect; the compressed image quality and
the compression rate are not only relational to the filter length, but also concerns with regularity
and local frequency, vanishing moment, orthogonality, bi-orthogonality. In this paper, we
58

The International Journal of Multimedia & Its Applications (IJMA) Vol.7, No.2, April 2015

implement adaptive lifting scheme based upon wavelet decomposition. Then, with the help of
IABC algorithm, we find the best directional window size to get better compression ratio with
considerable quality.

A. The Lifting Concept

Lifting is a spatial (or time) domain construction of bi-orthogonal wavelets. The lifting scheme
procedure consists of three steps: Split, Predict and Update (Fig. 1) and inverse Lifting scheme is
shown in Fig. 2.
Split:
Split the original data into two disjoint subsets. Though any disjoint split is possible, in the
standard lifting scheme we split the original data set x[n] into the even indexed points, xe[n]x[2n], and the odd indexed points xo[n]=x[2n+1]
Predict:
Generate the detail signals d[n] as the predicting error using prediction operator P
Update:
To obtain scaling coefficients c[n] that represent a coarse approximation to the original signal
x[n] merge xe[n] and d[n]. This is accomplished by applying an U update operator to the wavelet
coefficients and adding to

Figure 1. Lifting stage: Split, Predict, Update

As long as for the inverse and forward transforms U and P are chosen, the original signal will be
perfectly reconstructed. The inverse lifting stage is shown in Fig.2.

59

The International Journal of Multimedia & Its Applications (IJMA) Vol.7, No.2, April 2015

Figure 2. Inverse lifting steps: undo the Update, undo the Predict, and Merge the even and odd samples

B. Adaptive lifting scheme

The adaptive lifting scheme is classical lifting modification. Fig.3 shows the adaptive update
lifting scheme followed by a fixed prediction. At each sample n, according to a decision function
D(x[n],y) an update operator is chosen. As in the classical and space-varying lifting, the critical
point is that D(x[n],y) depends on y, and it also depends on the sample being updated. The update
operator and addition are fixed, in the standard lifting scheme. The choice of addition and the
update operator depends on the information locally available within both the approximation
signal and the detail signal in the adaptive lifting scheme.

Figure 3. Adaptive update lifting scheme

According to the structure of lifting, Adaptive Lifting Scheme performs update first , and then
performs prediction.

4. WAVELET TRANSFORM
The wavelet transform decomposes a typical image data to a few coefficients with large
magnitude and many coefficients with small magnitude. In wavelet transform first step
decomposes a signal into constituent parts in the time-frequency domain on a basis function
localized in both time and frequency domains. The image or signal is decomposed into four
different frequencies: approximation, vertical detail, horizontal detail and diagonal detail.
60

The International Journal of Multimedia & Its Applications (IJMA) Vol.7, No.2, April 2015

Figure 4. 2-level Wavelet decomposition

Up to a level the decompositions are repeated on the approximation coefficients. As details are
not decomposed at the high levels and can be described by the small scale wavelet coefficients,
wavelet transform is not suitable for images having rapid variations.

5. ARTIFICIAL BEE COLONY ALGORITHM (ABC)

Karaboga [17] in 2005 proposed ABC algorithm and the performance of ABC is analyzed in
2007 [18]. The ABC algorithm is developed by inspecting the behaviours of the bees on finding
food source, called as nectar, and the information of food sources to the bees in the nest is shared.
The different phases of ABC are classified into three types, namely, the employed bee, onlooker
bee, and the scout bee. In the employed bee phase, the employed bee stays on a food source and
provides the neighbourhood of the source in its memory. In the onlooker phase, the onlooker
receives the information of food sources from the employed bees in the hive and to gather the
nectar one of the food source is selected; and for finding new food, the new nectar, sources the
scout is responsible . The process of the ABC algorithm is presented as follows:
Step 1. Initialization: In a solution space randomly Spray percentage of the populations, and
then their fitness values calculated, called as nectar amounts, where the ratio of employed bees to
the total population is represented by . The populations positioned into the solution space, are
called as employed bees.
Step 2. Move the Onlookers: The probability of selecting a food source is calculated, select a
food source to move by roulette wheel selection for each onlooker bees and then determine the
nectar amounts of them.
Step 3. Move the Scouts: By a continuous predetermined number of iterations, if the fitness
values of the employed bees do not improve, which is called Limit, such food sources are
abandoned, and these employed bees become the scouts. The movement of scouts takes place.
Step 4. Update the Best Food Source Found So Far: Memorize the position and the best fitness
value, which are found by the bees.
Step 5. Termination Checking: Check whether the termination condition is satisfied by the
number of iterations. Terminate the program and output the results if the termination condition is
satisfied; otherwise go back to the Step 2.
61

The International Journal of Multimedia & Its Applications (IJMA) Vol.7, No.2, April 2015

6. INTERACTIVE ARTIFICIAL BEE COLONY ALGORITHM (IABC)

In IABC algorithm, in the solution space randomly spray percentage of the populations , then
fitness values called as nectar amounts is calculated, which represents the ratio of employed bees
to the total population. When these populations are positioned into the solution space they are
called employee bees. The probability of selecting a food source is then calculated. Select a food
source to move by roulette wheel selection for every onlooker bees and then nectar amounts of
them is determined. If the employed bees fitness values does not improve by predetermined
number of iterations continuously, called Limit, such food sources are abandoned, and these
employed bees become the scouts. The scouts are moved. The position and the best fitness value
found by the bees are memorized. We check whether the termination condition is satisfied by the
total number of iterations. If the condition for termination is satisfied, terminate the program and
output the results. The flow chart for IABC is shown in Figure 5.
In the proposed method, the input image is decomposed using wavelet lifting scheme and then
the Interactive artificial bee colony algorithm is used in the update process to get considerable
quality.
A. Algorithm
Step 1: Input the Image.
Step 2: Split the image into odd and even pixel regions.
Step 3: Decompose the image as (odd-even) for next prediction step.
Step 4: Fix the maximum coverage size as M and initialized K=0 for prediction co-efficient.
Where M is maximum window size, upto which it will do local search for each center
pixels maximum window size in our program is 5.
Step 5: Each pixel in the decomposed image is Scanned and its present fitness value and
compression ratio is calculated.
Step 6: To predict a and b call direction finding algorithm co-efficient of all 8-direction
combination.

62

The International Journal of Multimedia & Its Applications (IJMA) Vol.7, No.2, April 2015

Figure 6. Directional coefficient for center pixel x

63

The International Journal of Multimedia & Its Applications (IJMA) Vol.7, No.2, April 2015

Where x is an center pixel to be update

HL is predicted coefficient in horizontal left direction
HR is predicted coefficient in horizontal right direction
VT is predicted coefficient in vertically top direction
VD is predicted coefficient in vertically down direction
DL is predicted coefficient in diagonally top left direction
DR is predicted coefficient in diagonally top right direction
DB is predicted coefficient in diagonally bottom left direction
DD is predicted coefficient in diagonally bottom right direction
Step 7: By using Update lifting formula for each direction prediction calculate update weight and
find compression ratio and PSNR.
Step 8: The best individual is memorized , CR and its direction using IABC local search.
Step 9: To predict and update the best value for different range of window size iterate K
from (0 to M) .
Step 10: Using IABC local search, memorize the best window size in terms of its MSE and CR
for each reference pixel.

7. EXPERIMENTAL RESULTS
The IABC algorithm is tested on standard images with different image formats such as tif, png
jpeg. The results obtained using IABC are shown in figure 7. The results are tabulated for various
images in Table(1) & Table(2). Figure 8 & Figure 9 respectively presents the result for
Cameraman Image & Pepper Image graphically. The results obtained are with the results of
different image compression scheme such as wavelet transform, Artificial Bee colony algorithm
& Interactive Artificial Bee colony algorithm.
Image 1: Cameraman Image

Image 2: Peppers Image

(a)
(b)
Figure 7. Reconstructed Images with Lifting using IABC (a) Original Image (b) Output Image.
64

The International Journal of Multimedia & Its Applications (IJMA) Vol.7, No.2, April 2015

A. Comparison Tables
Table 1: Comparison table for cameraman image

Table 2: Comparison table for Pepper Image

65

The International Journal of Multimedia & Its Applications (IJMA) Vol.7, No.2, April 2015

A. Comparison Graphs

Figure 8. Comparison Graph for Cameraman Image

Figure 9. Comparison Graph for Pepper Image

8. CONCLUSION AND FUTURE ENHANCEMENTS

In this paper, Interactive Artificial Bee Colony algorithm is implemented in update process of
lifting scheme to give better PSNR. From the experimental results, it is concluded that IABC
method gives best results compared to other techniques like wavelet and Artificial Bee Colony
method. It gives the way to reduces in data to represent the image and decrease in transmission
bandwidth. So, the transmission cost and memory cost is reduced with help of IABC method.
In future work, Interactive Artificial Bee Colony algorithm can be implemented in the
thresholding process to reduce the number of coefficient representing the image by optimally
choosing the thresholding value to get more better compression and quality.

REFERENCES
[1]
[2]

Subramanya A. Image Compression Technique, potentials IEEE, Vol. 20, issue 1,pp19-23, FebMarch 2001.
Rafael C. Gonzalez, Richard Eugene; Digital image processing, Edition 3, 2008,page 466.
66

The International Journal of Multimedia & Its Applications (IJMA) Vol.7, No.2, April 2015
[3]
[4]
[5]
[6]

[7]

[8]

[9]
[10]
[11]

[12]
[13]
[14]
[15]

[16]

[17]
[18]
[19]
[20]
[21]

[22]

[23]

W. Sweldens, The lifting scheme: A new philosophy in biorthogonal wavelet constructions, in

Proc. SPIE, vol. 2569,1995,pp. 68-79.
W. Sweldens, The lifting scheme: A construction of second-generation wavelets, SIAM J. Math.
Anal., vol. 29, no. 2, pp. 511-546,1997.
A.R. Calderbank, I Daubechies, W. Sweldens, and B-L Yeo, Wavelet transforms that map integers
to integers, J. Appl. Comput. Harmon, Anal., vol.5, no. 3, 1998.
M. Adams and F. Kossentini, Reversible Integer-to-Integer Wavelet Transforms for Image
Compression: Performance Evaluation and Analysis, IEEE Trans.On Image Processing, vol.9, no.6,
pp. 1010-1024, Jun 2000.
N.V Boulgouris, D. Tzovaras, and M.G. Strintzis, Lossless image compression based on optimal
prediction, adaptive lifting, and conditional arithmetic coding, IEEE, Trans. Image Process., vol. 10,
no. 1,pp. 1-14, Jan. 2001.
M. Dorigo and L.M. Gambardella, Ant Colony Optimization for Data Clustering, Proc. Of 8th
Pacific Rim International Conference on Artificial Intelligence, Auckland, New Zealand, LNAI 3157,
pp. 534-543,2004.
J. Kennedy, R.C. Eberhart, and Y.Shi, Swarm Intelligence, Morgan Kaufmann Publishers, San
Francisco,2001.
B. Akay and D. Karaboga, Parameter tuning for the artificial bee colony algorithm, ICCCI 2009
(R.Kowalezyk, N.T. Nguyen and S.M.Chen,eds.), LNAI, vol. 5796, 2009, pp. 608-619.
V . U Kale and N.N. Khalsa, Performance evaluation of various wavelets for image compression of
natural and artificial images, International Journal of Computer Science and communication
1(2010), no.1, pp. 179-184.
D. Karaboga and B. Akay, A survey: Algorithms simulating bee swarm intelligence, Artificial
Intelligence Review 31(2009), no. 1, pp. 55-68.
G. Piella and H.J.A.M. Heijmans, Adaptive lifting schemes with perfect reconstruction, Research
Report PNARO104,CWI, Amsterdam, Feb. 2001.
F.W. Moore, A genetic algorithm for optimized reconstruction of quantized signals, Evolutionary
computation, 2005. The 2005 IEEE congress on, vol. 1,2005, pp. 105-111.
R. Ramanathan, K. Kalaiarasi, D. Prabha, Improved wavelet based compression with adaptive lifting
scheme using Artificial Bee Colony algorithm, International Journal of Advanced Research in
Computer Engineering & Technology, Vol 2 , Issue 4, April 2013.
M. Mohamed Ismail, Dr. K. Baskaran, Adaptive Lifting Based Image Compression Scheme Using
Artificial Bee Colony Algorithm, International Journal of Electronics Communication and Computer
Engineering, Vol. 4, Issue 1, ISSN(print):2278-4209, 2013.
W. Trappe and K.J.R.Liu, Adaptivity in the lifting scheme, in 33th Conference on Information
science and systems, Baltimore, March 1999, pp. 950-958.
D. Karaboga, An Idea Based On Honey Bee Swarm For Numerical Optimization, Technical
Report-TR06, Erciyes University, Engineering Faculty, Computer Engineering Department, 2005.
D. Karaboga and B.Basturk. On the performance of artificial bee colony (ABC) algorithm, Applied
Soft Computing 8(2008), pp. 687-697, 2008
E. Bonabeau, M. Dorigo, and G. Theraulaz, Swarm Intelligence: From Natural to Artificial
Intelligence, NY: Oxford University Press,1999.
Richa Goyal, Jasmeen Jaura, A Review of Various Image Compression Techniques, International
Journal of Advanced Research in Computer Science and Software Engineering, Vol. 4, Issue 7, July
2014.
Y. Guo, X.Gao, H. Yin, and Z. Tang, Coevolutionary Optimization Algorithm with Dynamic Subpopulation Size, International Journal of Innovative Computing, Information and Control, vol.3,
no.2, pp.435-448, 2007.
Pei-Wei Tsai, Jeng-Shyang Pan, Bin-Yih Liao, and Shu- Chuan Chu, Enhanced Artificial Bee
Colony Optimization, International Journal of Innovative Computing, Information and control,
vol.5, no.12, pp.1-12, Dec. 2009.

67

The International Journal of Multimedia & Its Applications (IJMA) Vol.7, No.2, April 2015
[24] Vrinda Shivashetty, G.G Rajput, Adaptive Lifting Based Image Compression Scheme Using
Interactive Artificial Bee Colony Algorithm, CSITY, SIGPRO, DTMN-2015, pp. 09-21, DOI:
10.5121/csit.2015.50302
[25] Dalijit Singh, Sukhjeet Kaur Ranade, Comparative Analysis of Transform based Lossy Image
Compression Techniques, International Journal of Engineering Research and Applications(IJERA) ,
Vol.2, Issue 5, September-October 2012, pp. 1736-1741.

AUTHORS
Dr. G. G. Rajput currently working as Associate Professor in the Department of Computer Science at Rani
Channamma University Belagavi, Karnataka State, India. The Area of interest includes Image processing,
Pattern recognition, Operations Research and Software Engineering.
Vrinda Shiva Shetty received B.E from Gurbarga University and M.Tech degree from VTU in Computer
Science and Engineering and presently pursuing Ph.D in Image Compression from the University of
Gulbarga University, Gulbarga, and Karnataka. Field of Interest includes Intelligent Image Processing,
Evolutionary Computation.

68