International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169

Volume: 4 Issue: 4 638 - 643

_______________________________________________________________________________________________
An Image Compression Encryption Hybrid Algorithm Using DWT and Chaos
System
A. Ilakkiya1 Dr. M. Pushparani2
M.PHIL Scholar Dept. of Computer Science 1 Head Department of Computer Science
Mother Teresa Womens University, Madurai Mother Teresa Womens University,
[email protected] Madurai

Abstract Recent developments of digital image production and applications have increases importance of digital image compression and
security in todays world. The proposed method is developed to combine both compression and security of image. Compression is achieved by
the deletion of redundant data. Discrete Wavelet Transform (DWT) is a in recent times developed compression technique in image compression.
The existing methods to encrypt images usually treat the whole matrix as the key which makes the key too large to distribute and memorize or
store. To solve this problem, in this proposed method key matrix is constructed using the logistic diagram and the Arnold transform is used for
image position scrambling. Initially the original image is decomposed into bands and compressed by level dependent hard threshold technique
and then combined with above encryption algorithms to get compressed-encrypted image. This algorithm produces a cipher for the test image
that has good diffusion and confusion properties. Simulation results of the histogram analysis, key sensitivity analysis of adjacent pixels, PSNR,
are representing the scrambling effect, security of the proposed algorithm and considerable compression performance.

Keywords- image compression, encryption, Arnold transform, chaos system.

__________________________________________________*****_________________________________________________

Introduction Industries, Retail Stores, Federal Government Agencies,

Security Industries, Museums and Galleries etc.
During the last decade, the use of computer networks has The goal of image compression is to decrease the amount of
grown-up enormously, and this growth continues unabated. data required to represent a digital image to decrease the large
Almost all networks are being installed, interconnected, and storage ability and transmission bandwidth. The Discrete
connected to the universal internet. Through internet more and wavelet transform (DWT) has gained well-known recognition
more information has been transmitted. The information is not in signal processing and image compression. The presentation
only text also audio, image, and other multimedia. As the of DWT based coding depends on the wavelet decomposition
developing of networks and multimedia communications level and threshold value.
technology, the problems of digital multimedia information
B. DWT Technique Arnold transformation
security and intellectual property protection and authentication
issues become increasingly prominent. For example, it is vital Wavelet analysis [2, 4, 6] can be used divided the information
to protect the diagrams of army emplacements, the diagrams of an image into approximation and detailed subordinate
of bank building construction, and the important data captured signal [4]. The approximation subordinate signal shows the
by military satellites. In addition, the number of computer general trend of pixel value, and three detailed subordinate
crimes has improved recently. Image security has become an signal show vertical, horizontal and diagonal details or
important topic in the modern computer world. To solve those changes in image. If these detail is very small than they can be
problems, the common method is image scrambling set to zero without considerably changing the image. If the
technology. In recent years, many researches of watermark number of zeroes is larger than the compression ratio is also
preprocessing only confined to the location scrambling, and greater. There is two types of wavelet is used. First one is
didnt guarantee the security. This paper puts forward an Continues wavelet transform[2] and second one is Discrete
encryption algorithm combing Logistic chaos system and wavelet transform.[2] Wavelet analysis is computed by filter
position scrambling system (Arnold transform), and reach a bank. There is two type of filter
better effect. It can enhance the robustness of image 1) High pass filter: high frequency information is kept, low
encryption. frequency information is lost.
A. Compression
2) Low pass filter: low frequency information is kept, high
In modern day, many applications need large number of frequency information is lost.
images for solving problems. Digital image [2] can be store on
disk. This storing space of image is also important. Because So signal is successfully decomposed into two parts, a detailed
less memory space means less time required to processing for part (high frequency) and approximation part(low frequency).
image. Here the concept of image compression comes. Image Level 1 is approximation, Level2, 3, &4 are correspondingly
compression [2] means reduced the amount of data essential to horizontal, vertical and diagonal of the image signal.
represent a digital image. There are lot of applications [3]
where the image compression is used to effectively increased
efficiency and performance. Applications are like Health

638
IJRITCC | April 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 4 Issue: 4 638 - 643
_______________________________________________________________________________________________
Fig.2. showing how the linear diagram stretches the unit
square and how its pieces are rearranged when the modulo
action is performed. The lines with the arrows show the
direction of the toning and increasing eigen spaces.
II. THE PROPOSED IMAGE COMPRESSIONENCRYPTION
ALGORITHM

A. Key matrix generation

The Key matrix is constructed as a circulate matrix. The
original row vector of the circulate matrix is controlled by the
logistic chaos map. The steps are as follows [1]:

1. A sequence with length 2N by logistic map with

initial condition X01 is generated; abandon the
Fig.1.Two-level wavelet analysis preceding N elements to obtain the sequence, which
C. Threshold coding Method are used as the initial row vectors of the circulate
matrices.
In level dependent threshold coding method, each transform
coefficient is compared with a threshold. If it is smaller than 2. The circulate matrix is constructed with the initial
the threshold then it is set to zero. If it is larger then it will be row vectors. To reduce the relevance among the
retained. Different threshold values for different column vectors, the first element of vector will be
decomposition level are used. By applying hard threshold the the result of multiplying by , where 2 < i < M and
coefficients lower this threshold level are zero, and the output > 1, and the iteration:
after a hard threshold is functional and defined by this
equation:

(1) Image compression encryption algorithm

The proposed algorithm is suitable for the image whose width
Where x(t), the input signal and T is the threshold. equals to height like the size of input image is N N. The
proposed method is shown in Fig.3a and b, and the image
D. Logistic map compression encryption steps are as follows:
Chaos system is often used in cryptography due to its pseudo
randomness and sensibility to the initial condition, the 1 Decompose the original image into 4 bands, band 1
definition of Logistic map is (lower), band 2, 3 and 4(higher), as shown in Fig.3a,
thus each band is of the size (N/2, N/2) using DWT.
(2) 2 Generate the compressed image by applying level
dependent hard threshold technique.
It becomes chaotic when the parameter . 3 Construct two (N/2 N/2) key matrices, and ,
E. Arnold transformation with keys X01 and X02, and multiple with the bands as
shown in Fig.3. C1, C2, C3 and C4 are the encrypted
Arnold's Cat Map transformation applied to an image to
matrices corresponding to band1, band2, band3 and
randomly rearrange the pixels of the image. We can define
band4.
Arnold transformation as follows. Let (x,y) is pointing in the
4 Scramble the bands by the Arnold transform with
unit square. Its move to (x,y) by the following equation and n
iteration1 and iteration 2 as show in Fig.3a. A1, A2,
is the order of the image.
A3 and A4 are the scrambled matrices corresponding
to band1, band2, band3 and band4. Image A is the
compressed encrypted image
The decryption and reconstruction process are shown in
Fig.3b. First the inverse Arnold transform performed and then
the key removed to produce the image bands. Next Inverse
DWT is applied to produce the reconstructed image.
III. SYSTEM MODELS
The system model contains 4 main modules.They are Input,
DWT, Key Matrix Generation,Arnold transform,Encrypted
Fig.2. The linear map stretches the unit square and its pieces rearrangement Image.

(3)

639
IJRITCC | April 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 4 Issue: 4 638 - 643
_______________________________________________________________________________________________
A. INPUT IMAGE B. DWT
A discrete wavelet transform is used to decompose
the image.

Fig.4.1. INPUT IMAGE

Fig.4.2. DWT IMAGE

Fig.3.a. Process flow of the encryption algorithm

Fig.3.b. Process flow of the decryption algorithm

C.

640
IJRITCC | April 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 4 Issue: 4 638 - 643
_______________________________________________________________________________________________
IV. EXPERIMENTAL ANALYSIS AND RESULTS
D. KEY MATRIX GENERATION
Chaos system is used to create key matrix for encryption.

Fig.6. (a) Encrypted Image using X01 = 0.11 and X02 = 0.23; (b) encrypted
Image using X01 + = 0.11 + 1 1016 and X02 = 0.23; and (c) difference
between two encrypted images (a) and (b).

Table 1
Correlation coefficients of adjacent pixels.
[1] Correlation
[2] Horizontal [3] Vertical [4] Diagonal
coefficient
Fig.4.3. MEASUREMENT MATRIX
[5] Input image [6] 0.9590 [7] 0.9217 [8] 0.9071
E. ARNOLD TRANSFORM [9] Existing [10] 0.0846 [11] 0.0583 [12] 0.0931
The pixels of the blocks are scrambled by Arnold method
transform. [13] Proposed [14] 0.0624 [15] 0.0108 [16] 0.0608
method
[17] Input image [18] 0.9585 [19] 0.9529 [20] 0.9064
[21] Existing [22] 0.0639 [23] 0.0539 [24] 0.0848
method
[25] Proposed [26] 0.0323 [27] 0.0274 [28] 0.0527
method

Fig.4.4. ARNOLD TRANSFORM IMAGE

F. OUTPUT IMAGE

Fig.4.5.(a). Input Image Fig. (b). Compressed Image

Fig.(c). Encrypted Image Fig.(d). Reconstructed Image

Fig.7. Correlation distribution of two horizontally adjacent pixels in (a1)

original Lena; (a2) encrypted Lena; (b1) original Cameraman; (b2) encrypted
Cameraman; (c1) original Peppers; and (c2) encrypted Peppers.

641
IJRITCC | April 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 4 Issue: 4 638 - 643
_______________________________________________________________________________________________
A. Histogram uniform in distribution or the second best when the histograms
of different encrypted images are similar to each other. Fig.
The image histogram is often used to analyze the performance 5(a1), (b1) and (c1) are the histograms of Lena, Cameraman
of the image encryption algorithm. It is the best when the and Peppers, respectively. And Fig. 5(a2), (b2) and (c2) are
values in the histogram of the encrypted image are fairly the histograms of their encrypted images, correspondingly.

Fig.5. Histogram: (a1) Lena; (a2) encrypted Lena; (b1) Cameraman; (b2) encrypted Cameraman; (c1) Peppers; and (c2) encrypted Peppers.

The key space for X01 is equal to (1/ ), where is the value
The histograms of the two original images are obviously
different from each other, while their encrypted images have of for MAE = 0. The simulation results show that comes
similar histograms. After a large number of parallel
out to be 1 10 , i.e., the key space of X01 is 1 1017.
17
experiments, shows that the histograms of the cipher texts of
Similarly, the key space of X02 is 1 1017. Thus, the total key
different original images are similar to Fig. 5(a2), (b2) and
space is as large as 1085. If one wants to construct the correct
(c2). That is to say, the proposed algorithm can frustrate the
measurement matrix by exhausting the keys, she must
statistical analysis attack.
calculate 1085 times which would take much time. Thus the
B. Key space proposed algorithm is secure against brute-force attack.
An image encryption algorithm should be secure even though
everything is known except for the key. Thus a good C. Correlation of two adjacent pixels
encryption algorithm should have a large enough key space. In The correlation of two adjacent pixels in a meaningful image
the proposed algorithm, X01 and X02 are used as keys. Here, ~ is usually close to 1, while that of the encrypted image should
the key space is calculated for X01 as generate two different be close to 0. To measure the correlations between two
sequences and by using X01 and X01 + as initial values adjacent pixels in the horizontal, vertical and diagonal
and both sequences are of length N, and dene mean absolute directions, 16,000 adjacent pixel pairs are selected randomly
error between the two sequences as [14] from original image and encrypted image, correspondingly.
(5) original image and the disordered distribution reflects the
642
IJRITCC | April 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 4 Issue: 4 638 - 643
_______________________________________________________________________________________________
weak correlation between two adjacent pixels in the encrypted [4] Swastik Das and Rashmi Ranjan Sethy, A Thesis on
image. And the distributions of vertical and diagonal Image Compression using Discrete Cosine Transform and
directions share the similar modality. The correlation Discrete Wavelet Transform, Guided By: Prof. R.
coefficient is Baliarsingh, dept of Computer Science & Engineering,
National Institute of Rourkela.
[5] Pavan Kumar Goswami, Namita Tiwari, Meenu Chawla,
(6) Block Based Image Encryption Using Iterative Arnold
Transformation, International Journal of Advanced
Where and . Research in Computer Science and Software Engineering
The quantitative evaluation on correlation is compiled in Table 3(8), August - 2013, pp. 273-278
1. That shows that the proposed algorithm removes the tight [6] Karen Lees Image compression using Wavelets, Report
relationship between adjacent pixels of the original image of M.S. 2002.
successfully. The results demonstrate that the attackers cannot [7] B. Bhargava, C. Shi, and S. Y.Wang, MPEG video
obtain useful information by statistical analysis and the encryption algorithms, Multimedia Tools and
proposed algorithm can resist statistical analysis. Applications, vol. 24, no. 1, pp. 5779, 2004.
analysis the mean square error (MSE) [13] between decrypted [8] A. Cohen and J. Kovacevic, Wavelets: the mathematical
image and original image is calculated as background, Proceedings of the IEEE, vol. 84, no. 4, pp.
514522, 1996.
(7)
[9] Wangsheng Fang1, Lulu Wu1,Rong Zhang1, A
where M N represents the total number of image pixels, I(x, Watermark Preprocessing Algorithm Based on Arnold
y) and D(x, y) denote the values of input image and output Transformation and Logistic Chaotic Map, Advanced
image at the pixel (x , y), respectively. The sensitivity can also Materials Research Vols. 341-342 (2012) pp 720-724
be tested by comparing two encrypted images obtained by [10] Pavan Kumar Goswami, Namita Tiwari, Meenu Chawla,
using neighbor keys. Fig. 7(a) and (b) shows the encrypted its Block Based Image Encryption Using Iterative Arnold
Transformation, International Journal of Advanced
neighbor keys X01 = 0.21 + 1 1016, X02 = 0.35, respectively.
Research in Computer Science and Software Engineering
And the difference between these two encrypted images is 3(8), August - 2013, pp. 273-278
shown in Fig. 7(c). The tiny change in the keys results in great
[11] Lu P, Xu ZY, Lu X, Liu XY. Digital image information
changes in the encrypted image. From the above two cases, the encryption based on compressive sensing and double
proposed algorithm is sensitive enough to the keys. random-phase encoding technique. Optik-Int J Light
Electron Opt 2013;124:25148.
V. CONCLUSION
[12] Takanori N, Bahram J. Optical encryption system with a
In this paper a hybrid algorithm for compression-encryption binary key code. Appl Opt 2000;39:47837.
has been proposed which based on key matrix and chaos [13] Chu Hui Lee and Zheng Wei Zhou, Comparison of Image
system with Arnold transform. The image is decomposed into Fusion based on DCT-STD and DWT-STD, IMECS Vol I,
4 bands to compress and encrypt. Then the compressed image March 14-15, 2012, Kong Kong.
is encrypted by chaos system with Arnold transform. By using [14] Hennelly BM, Sheridan JT. Image encryption and the
the chaos system the original key circulant matrix is fractional Fourier trans-form. Proc SPIE Int Soc Opt Eng
constructed and controlled, the proposed method is secure. By 2003;5202:7687
using Arnold transform the compressed and encrypted image
is scrambled, the security is enhanced further. The encrypted
image is analysised from histogram, key sensitivity, PSNR and
correlation of adjacent pixels, which shows that the proposed
method resists to different attacks. Since this algorithm uses
the number of Arnold transformation and the initial value of
Logistic chaos system as the key, the key space is big, and has
a strong sensitivity. From the simulation results the proposed
algorithm is secure and can provide good compression.

VI. REFERENCES
[1] Nanrun Zhou, Aidi Zhang, Fen Zheng, Lihua Gong Novel
image compressionencryption hybrid algorithm based on
key-controlled measurement matrix in compressive
sensing, Optics & Laser Technology 62 (2014)152160 .
[2] Rafael C. Gonzalez, Richard E. Woods. (1992), Digital
Image Processing(2nd edition), NJ:Prentice Hall
[3] Locker Gnome (2011), Real World Application Of Image
Compression,
http://www.lockergnome.com/nexus/windows/2006/12/25/r
eal-world-applications-of- image compression/ [accesed 11
Dec 2011].

643
IJRITCC | April 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________