ISSN : 2230-7109 (Online) | ISSN : 2230-9543 (Print) IJECT Vol.

3, IssuE 4, oCT - DEC 2012

Modified scheme for Lossy Compression and Iterative

Reconstruction for Encrypted Image
1
Asha P. Ghodake, 2Sujata Mendgudle
1
Bharati Vidyapeeth College of Engineering, Navi Mumbai, India
2
Ramrao Adik Institute of Technology, Navi Mumbai, India

Abstract image encryption, unencrypted plain-text image Xn is described

This paper presents advanced scheme for lossy compression of an by n bits a key sequence Kn of the same dimension as the plain-
encrypted image with flexible compression ratio. Encryption of text, consisting of bits selected uniformly at random. The image
original image is obtained by using pseudorandom permutation. is encrypted by applying a bitwise exclusive-OR (XOR) between
Then encrypted image is compressed by using orthogonal each key bit and the corresponding plain-text bit to produce the
transform and then the excessively rough and fine information cipher-text Yn = Xn ⊕ Kn. In this system, standard stream-cipher
in the coefficients is removed, leading to a reduced data amount. encryption and LDPC codes are used for compression [4].
After receiving the compressed image, receiver can reconstruct
the principle contents of original image by iteratively updating the II. System Overview
values of coefficients. For this process flexible compression ratio is Encrypt an original image using pseudorandom permutation, and
used. The compression ratio and the qualities of the reconstructed then encrypted data are efficiently compressed by discarding the
image are dependent on the values of compression parameters. excessively rough and fine information of coefficients generated
Higher the compression ratio and the smoother the original image, from orthogonal transform. A receiver can reconstruct the principal
the better the quality of the reconstructed image. The proposed content of the original image by iteratively updating the values
scheme significantly improves the compression efficiency. of coefficients.

Keywords A. Image Encryption

Lossy Compression, Orthogonal Transform, Compression Ratio, Image can be viewed as an arrangement of bits, pixels and
Encryption blocks. The intelligible information present in an image is due
to the correlations among the bits, pixels and blocks in a given
I. Introduction arrangement. This perceivable information can be reduced by
Now a day’s, large files such as digital images to be easily decreasing the correlation among the bits, pixels and blocks using
transmitted over the internet. The security of digital images certain permutation techniques. In this method, the pixel values
involves several different aspects. Encryption is used for content are same after encryption but their position will be changed. Image
confidentiality and access control through which only authorized and video encryption have applications in various fields [5]. The
parties holding decryption keys can access content in clear text. original image is in uncompressed format and each pixel with a
The objective of image compression is to reduce irrelevance and gray value falling into [0, 255] is represented by 8 bits. Denote
redundancy of the image data in order to be able to store or transmit the numbers of the rows and the columns in the original image as
data in an efficient form [1]. N1 and N2, and the number of all pixels as N = N1 x N2. Then,
the amount of bits of the original image is N x 8.
Number of pixels = N = N1 x N2
Number of bit = N x 8

B. Compression of Encrypted Image

The objective of image compression is to reduce irrelevance
and redundancy of the image data in order to be able to store or
transmit data in an efficient form. Image compression may be
lossy or lossless.
Lossy methods are especially suitable for natural images such as
photographs, compress multimedia data (audio, video, and still
Fig. 1: Secure Transmission of Redundant Data images). In the proposed scheme, a majority of pixels are converted
to a series of coefficients using an orthogonal transform, and then
A pseudorandom permutation is used to encrypt an original the excessively rough and fine information in the coefficients
image. In this, pixel position is changed and the encrypted data is removed, leading to a reduced data amount. In orthogonal
are efficiently compressed by orthogonal transform and discarding transform convert image pixel values to transform coefficient
the excessively rough and fine information. After receiving the values. This is a linear process and no information is lost, the
compressed data, reconstruct the principal content of the original number of coefficients produced is equal to the number of pixels
image by iteratively updating the values of coefficients. This way, transformed.
the higher the compression ratio and the smoother the original
image, the better the quality of the reconstructed image [2]. C. Steps for Compression of Encrypted Image
There are different methods of encryption and compression. The 1. The network provider divides Permuted pixel sequence, into
traditional way of securely and efficiently transmitting redundant two parts: the first part made up of (α. N) pixels and the second
data is to first compress the data to reduce the redundancy, and one containing the rest of the (1 - α).N pixels. The pixels in the
then to encrypt the compressed data [3]. In existing system for first part are called rigid pixels and the pixels in the second part

w w w. i j e c t. o r g InternatIonal Journal of electronIcs & communIcatIon technology 369

IJECT Vol. 3, IssuE 4, oCT - DEC 2012 ISSN : 2230-7109 (Online) | ISSN : 2230-9543 (Print)

are elastic pixels. It is called decomposition. Here, with the knowledge of M and L1, the receiver may calculate
2. Pixels in first part as p1, p2, p3,…………,p(α. N) and pixels in L2, and then get the values of Sk by converting binary blocks with
second part as q1, q2,…,q(1 - α ). N L2 bits into digit pieces in an M-ary notational system. Obtain Sk
The value of α is within (0, 1). and all the parameters from compressed data.
3. Perform an orthogonal transform in the elastic pixels to calculate 2. According to the secret key, the receiver can retrieve the
the coefficients positions of rigid pixels. That means the original gray values
Q1, Q2……… Q (1 - α). N at the positions, which distribute over the entire image, can be
[Q1, Q2……… Q(1 - α). N] = [q1, q2……………….q(1 - α). N]. H exactly recovered.
Here, H is a public orthogonal matrix with a size of (1 - α). N x 3. For the pixels at other positions, i.e., the elastic pixels, their
(1 - α).N and it can be generated from orthogonal zing a random values are firstly estimated as the values of rigid pixels nearest to
matrix. them. That means, for each elastic pixel, we find the nearest rigid
Size of H = (1 - α). N x (1 - α). N pixel and regard the value of the rigid pixel as the estimated value
4. For each coefficient Q, calculate of the elastic pixel. If there are several nearest rigid pixels with the
same distance, regard their average value as the estimated value of
the elastic pixel. Because of spatial correlation in the natural image,
k = 1, 2 … (1 - α). N the estimated values are similar to the corresponding original
Where ∆ and M are system parameters. values. In the following, the estimation will be iteratively updated
The round operation returns the nearest integer and the mod by exploiting the information of Sk.
operation gets the remainder. Qk is converted into an integer Sk 4. Rearrange the estimated values of elastic pixels using the same
within permutation way, and denote them as q’1 q’2 ……….. q’(1 - α). N
[0, M -1]. 5. Calculate the coefficients
5. Then Qk = rk. ∆ + Sk. (∆/M) + tk [ Q’1 , Q’2 ……… Q’(1 - α). N ]
rk = rough information and = [ q’1 q’2 ……….. q’(1 - α). N ] . H
tk = fine information 6. Calculate the differences
The rough information rk and the fine information tk are discarded.
While only the information Sk on the medium level remains.
The rough information rk will be retrieved by an iterative image k = 1,2,………….. (1 - α). N
reconstruction procedure, and the loss of the fine information tk 7. Modify the coefficients to the closest values consistent with
cannot seriously affect the quality of the reconstructed image, the corresponding Sk.
Where 8. Perform inverse transform
[ q’’1 q’’2 ……….. q’’(1 - α). N]
= [ Q’’1 , Q’’2 ……… Q’’(1 - α). N ] . H-1
9. Calculate the average energy of difference between the two
versions of elastic pixels
6. Segment the set of sk into many pieces with L1 digits and
calculate the decimal value of each digit piece. Then, convert each
decimal value into L2 bits in a binary notational system, • If D is greater than a threshold T (recommended 0.05), go
where back to step 5 –iterating.
L2= [L1.Log2 M] • Otherwise, terminate the iteration and output the image made
The total length of bits generated from all pieces of sk is up of the rigid pixels and the final version of elastic pixels.
L=(1-α).N.Log2 M
Then each value of sk is represented by Log2 M bits. III. System Design
7. Collect the data of rigid pixels, the bits generated from all 1. In encryption the pixel values are same after encryption but
pieces of Sk, and the values of parameters including N1, N2, their position will be changed.
α, ∆, M and L1 to produce the compressed data of encrypted 2. Permuted pixels are divided in to rigid and elastic pixels.
image. Compression ratio R is a ratio between the amounts of the Perform orthogonal transform in elastic pixels. Remove rough and
compressed data and the original image data, is approximately fine information and keep Sk. Finally collecting all data calculate
compression ratio.
3. In reconstruction according to the secret key, the receiver can
retrieve the positions of rigid pixels.
4. For each elastic pixel, we find the nearest rigid pixel and
regard the value of the rigid pixel as the estimated value of the
elastic pixel. If there are several nearest rigid pixels with the same
D. Image Reconstruction distance, regard their average value as the estimated value of the
In this system, iterative reconstruction is used because iterative elastic pixel. Because of spatial correlation in the natural image,
approach includes improved insensitivity to noise and capability the estimated values are similar to the corresponding original
of reconstructing an optimal image in the case of incomplete data values.
with the compressed data and the secret key. 5. Modify the coefficients to the closest values consistent with
the corresponding Sk.
E. Steps of Image Reconstruction 7. Perform inverse transform.
1. Decompose the compressed data and obtain the gray values 8. Calculate the average energy of difference between the two
of rigid pixels, the values of all Sk, and the values of parameters. versions of elastic pixels.

370 InternatIonal Journal of electronIcs & communIcatIon technology w w w. i j e c t. o r g

ISSN : 2230-7109 (Online) | ISSN : 2230-9543 (Print) IJECT Vol. 3, IssuE 4, oCT - DEC 2012

Fig. 3: (a) Original image Lena, (b) its encrypted version, (c).
the medium reconstructed image from compressed data with less
PSNR, and (d) the final reconstructed image with more PSNR

By collecting all data, we have calculated compression ratio R

for original image Lena, as shown in Table 1.

Table 1: Compression ratio R in Reconstructed Image with

Different Parameters for Test Image Lena
α R
M=7 0.16 0.45
M=7 0.12 0.42
M=7 0.09 0.40
M=6 0.16 0.43
M=6 0.12 0.40
M=6 0.09 0.38
M=4 0.16 0.37
M=4 0.12 0.34
M=4 0.09 0.31

V. Conclusion
In this work, image encryption is obtained by changing the position
of pixels. Performing orthogonal transform in permuted pixels
and then removing rough and fine information compression is
Fig. 2: Total Procedure of Encryption, Compression and obtained. In iterative procedure in each step update the values of
Reconstruction coefficients and reconstruction of original principal content. The
compression ratio and the quality of reconstructed image vary with
IV. Result different values of compression parameters. In general higher the
The test image Lena, shown in fig. 3(a) sized 512 x 512 used as compression ratio and the smoother the original image, the better
the original image. After pixel permutation, the encrypted data of the quality of the reconstructed image.
the image were produced, fig. 3(b). For showing their disorder, the
encrypted pixel sequence is rearranged as a matrix with size of References
512 x 512. Then, we compressed the encrypted data with different [1] A. Mitra, Y. V. Subba Rao, S. R. M. Prasanna,“A New Image
values of and M and then calculate the compression ratio R with Encryption Approach using Combinational Permutation
different values ofand M. With the compressed data, the receiver Techniques”, International Journal of Electrical and Computer
can retrieve the original content by using the image reconstruction Engineering, 1, 2 pp. 127-131, 2006.
procedure, fig. 3(c) and fig. 3(d). In which all rigid pixels are [2] Xinpeng Zhang,“Lossy Compression and Iterative
recovered and the elastic pixels are estimated as the values of Reconstruction for Encrypted Image”, IEEE transactions
their nearest rigid pixels. on Information Forensics and Security, Vol. 6, No. 1, March
2011.

w w w. i j e c t. o r g InternatIonal Journal of electronIcs & communIcatIon technology 371

IJECT Vol. 3, IssuE 4, oCT - DEC 2012 ISSN : 2230-7109 (Online) | ISSN : 2230-9543 (Print)

[3] Wei Liu, Wenjun Zeng, Lina Dong, and Qiuming Yao
“Efficient Compression of Encrypted Grayscale Images”,
IEEE Transactions on Image Processing, Vol. 19, No. 4, pp.
10971102, April 2010.
[4] D. Schonberg, S. C. Draper, C. Yeo, K. Ramchandran, “Toward
compression of encrypted images and video sequences”,
IEEE Transactions Information Forensics Security, Vol. 3,
No. 4, pp. 749–762, Dec. 2008.
[5] Daniel Schonberg, Stark C. Draper, Chuohao Yeo, Kannan
Ramchandran,“Towards Compression of Encrypted Images
and Video Sequences”.
[6] Ibrahim Fathy El-Ashry,“Digital Image Encryption”, A Thesis
Submitted for The Degree of M. Sc. of Communications
Engineering.

Asha Pandit Ghodake received her B.E. degree in Electronics

from Shivaji University, Kolhapur, Maharshtra in 2005. Currently
she is persuing M.E. from University of Mumbai, Mumbai
under the guidance of Prof. Sujata Mendgudle. She is currently
working as Asst. Professor, with Department of Electronics &
Telecommunication Engineering, Bharati Vidyapeeth College of
Engineering, University of Mumbai, Mumbai from 2010. Earlier
she worked with SSPM’s College of Engineering, University of
Mumbai, Mumbai from 2007 to 2010. Her research interests
include digital signal processing and microwave.

Sujata Mendgudle received her B.E. degree in Electronics from

Shivaji University, Kolhapur, Maharshtra in 1998 and M.E. from
BATU, Lonere in year 2005. She is currently working as Asst.
Professor, with Department of Electronics Engineering, Ramrao
Adik Institute of Technology, University of Mumbai, Mumbai.
Her research interests include digital signal processing and
microwave.

372 InternatIonal Journal of electronIcs & communIcatIon technology w w w. i j e c t. o r g