0% found this document useful (0 votes)
25 views

Selective Image Encryption in Fractional Wavelet Domain

This document discusses a proposed technique for selective image encryption in the fractional wavelet domain. The technique aims to encrypt only significant subbands using a chaotic stream cipher, while overcoming limitations of previous techniques. It utilizes the relationship between subband normalized information energy and perceptual information to select significant subbands for encryption. Analysis shows the technique provides better perceptual and cryptographic security with less computational time compared to existing fractional wavelet domain encryption methods.

Uploaded by

Abid
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
25 views

Selective Image Encryption in Fractional Wavelet Domain

This document discusses a proposed technique for selective image encryption in the fractional wavelet domain. The technique aims to encrypt only significant subbands using a chaotic stream cipher, while overcoming limitations of previous techniques. It utilizes the relationship between subband normalized information energy and perceptual information to select significant subbands for encryption. Analysis shows the technique provides better perceptual and cryptographic security with less computational time compared to existing fractional wavelet domain encryption methods.

Uploaded by

Abid
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

Int. J. Electron. Commun.

(AEÜ) 65 (2011) 338–344

Contents lists available at ScienceDirect

International Journal of Electronics and


Communications (AEÜ)
journal homepage: www.elsevier.de/aeue

Selective image encryption in fractional wavelet domain


Nidhi Taneja a,∗ , Balasubramanian Raman b , Indra Gupta a
a
Department of Electrical Engineering, Indian Institute of Technology, Roorkee, India
b
Department of Mathematics, Indian Institute of Technology, Roorkee, India

a r t i c l e i n f o a b s t r a c t

Article history: Fractional wavelet domain inherits the virtues of wavelet and fractional domain, to provide improved
Received 23 September 2009 security. Non-consideration of data significance and emphasising on fractional order for the entire secu-
Accepted 5 April 2010 rity consumes high computational resources without any proportional effect on the security. This article
proposes an efficient selective encryption in fractional wavelet domain that encrypts only significant
Keywords: subbands using a chaotic stream cipher. Relationship between normalized information energy and per-
Arnold cat map
ceptual information of a subband is utilized to select the significant subbands. Thorough performance
Fractional wavelet transform
and security analysis reflects better perceptual and cryptographic security with less computational time.
Normalized information energy
Selective encryption
© 2010 Elsevier GmbH. All rights reserved.

1. Introduction improved security. This offers the fractional order as an extra key, in
addition to the keys offered by any wavelet domain based encryp-
Advancement in communication and networking has led to tion technique.
numerous image applications in day-to-day life. With an increase in Chen and Zhao [10] proposed image encryption in fractional
transmission and distribution of digital image data through open- wavelet transform (FRWT) domain, where fractional order of
natured wired or wireless IP networks, piracy operations have also FRFT and scaling factor of WT are used as the potential keys.
increased. To restrain these activities, security of digital data is However, considering the fractional order as the only encryp-
required at different stages of data archival, transmission and dis- tion key does not provide cryptographic security. This yields a
tribution. small key space, (0 − 2), due to the periodic property of frac-
Encryption is considered as the first line of defence in digital tional order, F 0 [f (t)] = F 2 [f (t)] = f (t). Due to this limited key
rights management security solutions. This provides data confiden- space, an approximate image can be obtained if the assumed frac-
tiality and consequently prevents illegal copying and distribution tional order is in proximity of the correct fractional order. This
of valuable data. Application specific total or selective encryption makes the encryption technique vulnerable to various cryptana-
techniques have been developed in different domains, with a major lytic attacks.
thrust on selective encryption that trade security for computational The attained security level is improved upon by encryption of
efficiency by encrypting only the most important part of multime- randomly selected subbands [11]. This consumes more compu-
dia data [1]. tational resources without any proportional improvement in the
As data significance can easily be determined in transforma- security level achieved by the encryption technique. This is due to
tion domain, it is a preferable domain for selective encryption of the fact that random selection of subbands may not always lead to
digital images. Several discrete cosine transform (DCT) domain significant subband selection.
and wavelet transform (WT) domain based encryption techniques Despite the superiority of fractional combined domain, initial
have been proposed that permutes or encrypts selected coefficients attempts of image encryption in this domain had issues relating
[2–4]. In addition to these well established domains, recent years to security keys, key space, perceptual degradation, etc. [11]. This
have also witnessed fractional Fourier transform (FRFT) domain as necessitates the development of an efficient and improved selective
a potential transformation domain for encryption [5–7]. encryption technique, that would overcome the limitations of the
Efforts have been made to integrate FRFT domain with wavelet existing techniques.
or wavelet packet domain [8,9]. This fractional combined domain This article develops a simple, yet efficient selective encryption
inherits the virtues of wavelet and fractional domain to provide technique that considers fractional order as a part of the key struc-
ture and not the entire security key. Chaos based Arnold cat map
is utilized to achieve data confidentiality, and at the same time, to
∗ Corresponding author. increase the effective key space. The subband selection has also
E-mail address: [email protected] (N. Taneja). been done on the basis of data significance, as opposed to ran-

1434-8411/$ – see front matter © 2010 Elsevier GmbH. All rights reserved.
doi:10.1016/j.aeue.2010.04.011
N. Taneja et al. / Int. J. Electron. Commun. (AEÜ) 65 (2011) 338–344 339

dom selection. Extensive experimental and comparative analysis formation matrix A is 1, the map is area preserving by nature.
ascertains the efficacy of the proposed technique. Iterated actions of A on a pixel form a dynamic system:

2. Preliminaries for proposed technique rn+1 = An r0 (modN)orrn+1 = Arn (modN) (4)

With each iteration of the Arnold cat map, the pixel value shifts
The proposed technique encrypts only few subbands selected on
to a new position. This results into a completely distorted matrix
the basis of their normalized information energy. The preliminaries
after a few iterations. Thus, making Arnold cat map, a preferable
for the selection and encryption of subbands in fractional wavelet
choice for image scrambling.
packet domain is discussed hence.

2.1. Fractional wavelet analysis 3. Proposed encryption technique

FRWT is a realization of the wavelet transform in FRFT domain The present section proposes a FRWT domain based encryption
[12,13]. FRFT has a unique property of describing the informa- technique. The schematic diagram of the proposed technique is
tion of spatial and frequency domain, due to the rotation of illustrated in Fig. 1, and the step-by-step procedure is explained
time–frequency plane over an arbitrary angle. In contrast, wavelet hereafter.
transform has a multiresolution property. A combination of these
two domains result into FRWT, that exhibits multiresolution 1. Perform ˛-order, l-level FRWT decomposition of the input
property, describing the spatial as well as frequency domain infor- image. Denote the various subbands as fl , where, 1 < l ≤ n,
mation.  ∈ {A, H, V, D}.
The mathematical representation for the FRWT of a one- 2. Calculate NIE of each subband and the entire transformed image
dimensional function f (t), having a fractional order 0 < ˛ < 2, is using Eq. (2). Denote these energies as El and Etotal , respectively.
given as follows: El denotes the energy of the subband fl , and Etotal denotes the

 total energy of the image.
W˛ (u, s, ) = F ˛ [f (t)](x)e−jux 3. To identify the significant subbands, set the threshold energy as
s, (x)dx (1)
Eth = Q × Etotal , where Q is the user-defined threshold factor and
−∞ 0 ≤ Q ≤ 1.
where s, , and denote the dilation parameter, translation param- 4. Identify the significant subbands by comparing the subband
eter, and the mother wavelet function, respectively. As compared energy with the threshold energy.
to wavelet domain, FRWT domain provides better encryption effi- 5. For a particular subband, if El < Eth , mark it as insignificant, and
ciency [11]. leave it unencrypted.
6. For a significant subband, i.e. El ≥ Eth , encrypt it as follows:
2.2. Normalized information energy
(a) Scramble the subband using n-iterations of the Arnold cat map.
Normalized information energy (NIE), for different subbands, in (b) Generate a random binary matrix B using a seed value b. The
a transformed image is expressed as follows: size of B is same as that of the scrambled subband.
M1 M2
(c) XOR the sign bits of the coefficients in the scrambled sub-
1 
2 band with the matrix B. This only modifies the sign bit of the
El = xi,j (2)
M1 × M2 coefficients and does not disturb their magnitude. This is math-
i=1 j=1

where xi,j is the value of transform coefficient at position (i, j),


1 ≤ l ≤ n, l denotes the decomposition level,  corresponds to
{A, H, V, D} and M1 × M2 is the size of the subband. Hence, El
denotes the energy of the chosen subband fl .
NIE is directly proportional to the visual information present in
a subband [14]. Thus, NIE can be employed to determine the signif-
icance of various subbands. Encryption of the selected significant
subbands would provide better security with less computational
time.

2.3. Arnold cat map

Arnold cat map [15] is a two dimensional map, that employs


shearing and wrapping operation to completely scramble a matrix.
Mathematically, Arnold cat map for a matrix of size N × N is
expressed as follows:
   
x′ x
=A mod(N) (3)
y′ y
 
1 p
where A = , p and q are positive integers, (x, y) ∈ [1, N],
q pq + 1
and det(A) = 1.
(x′ , y′ ) is the new position of original pixel position (x, y), when
the Arnold cat map is performed once. As determinant of the trans- Fig. 1. Schematic diagram for proposed encryption technique.
340 N. Taneja et al. / Int. J. Electron. Commun. (AEÜ) 65 (2011) 338–344

ematically represented as follows:

M(i, j) = A(i, j) ⊕ B(i, j) (5)

where A(i, j) denotes the sign bit of Arnold scrambled coeffi-


cient, B(i, j) denotes the bit from random binary matrix, and
M(i, j) denotes the modified sign bit.

7. Map the modified subbands to their original position, and take


the inverse FRWT to retrieve the encrypted image. Fig. 2. Encrypted output for test images.

The reverse process is applied at the receiver to decrypt the To evaluate the PSNR values, all parameters of the security key,
image. To perform decryption, the encrypted image and the secu- except the fractional order, are kept constant. This is performed to
rity keys are required at the receiver end. As it is a symmetric key observe the effect of fractional order on the encrypted output. The
cryptosystem, the encryption and decryption keys are same, and obtained PSNR values are indicated in Table 1. As PSNR of 28 dB is
expressed as (key: Q, ˛, n, b), where these four parameters denote the minimum threshold for perceptual similarity between any two
the threshold factor, fractional order, Arnold iterations and the seed images, the obtained PSNR values are compared with this value. It is
value, respectively. observed that the PSNR obtained for the test images, at all fractional
The encrypted image is first transformed into FRWT domain and orders, is less than 11 dB. This is well within the satisfactory limit.
subbands for the decryption process are selected using the same High perceptual degradation and low PSNR values obtained for the
NIE criterion. Sign bit decryption followed by arnold descrambling proposed technique reflects the satisfaction of the objective and
of selected subbands is performed. Inverse FRWT is then performed subjective evaluation metrics.
to retrieve the original image.
If correct keys are employed at the time of decryption, then 4.2. Statistical attack analysis
the decrypted image would be a replica of the original image. In
the case of an incorrect decryption key, the output may vary. The This attack is launched by exploiting the predictable relation-
next section analyzes the effect of key-sensitivity of the proposed ship between data segments of the original and the encrypted
technique. It also evaluates the performance and security of the image [16]. Histogram of an image reflects the distribution of pix-
proposed technique for different attack scenarios. els at different gray levels. Hence, this is analyzed to investigate the
existence of any relationship between the image-pair.
Histogram for the original and the encrypted image is indicated
4. Results and discussion in Fig. 3. A significant change in the gray level distribution of pixels
illustrates non-existent correlation between the two images. This
To assess the performance of the proposed technique, extensive does not leave enough scope to launch a statistical attack on the
analysis is performed on various grayscale images, using different proposed technique.
combinations of the security key. The values so obtained are an
indicative measure of the performance and security level attained
4.3. Key sensitivity analysis
by the proposed technique. For experimental purposes, the key
structure (Q, ˛, n, b) is fixed at (0.65, 1, 15, 6345921). Discussion A highly key sensitive encryption algorithm protects the
and conclusion are based on the analysis of numerous test images, encrypted data against various cryptanalytic attacks. While devel-
however visual results for only two images are shown here. In the oping a cryptosystem, it is assumed that an intruder knows the
entire analysis, the original image and its corresponding encrypted encryption structure and a-priori probability of the used key k ∈ K.
image are referred to as an ‘image-pair’. As per the Kerckhoff’s principle, only secrecy of the used key is
required. Even a strong or well designed cryptosystem can be bro-
4.1. Perceptual security and peak signal to noise ratio ken easily if the key is poorly chosen or the key space is too small.
This makes the encryption or decryption key as the most important
Subjective evaluation of the encrypted images is performed part of any cryptosystem. Thus, a good cryptosystem should satisfy
to assess the amount of information leakage. Fig. 2 shows the the following two conditions to verify the key sensitivity and key
encrypted output for the two test images, ‘Huts’ and ‘Mandril’. It space:
is observed that the obtained images are completely unintelligible,
and do not reveal any information about the original image. The • The key space should be discretized in such a way that two cipher-
degradation introduced is also objectively evaluated using PSNR texts encrypted by two slightly different keys k1 , k2 ∈ K should
[16]. be completely different.

Table 1
PSNR obtained for proposed technique.

Fr. order Crowd Barbara Lena Mandril Pillars Butterfly Alonehut Peppers

0.1 9.923 7.564 7.322 7.377 8.884 8.004 10.880 7.341


0.2 10.058 7.512 7.232 7.309 8.896 8.130 10.849 7.290
0.3 10.117 7.621 7.245 7.304 8.893 8.023 10.836 7.415
0.4 10.09 7.601 7.326 7.3645 8.878 8.097 10.892 7.363
0.5 9.408 6.975 6.686 6.613 8.222 7.475 10.214 6.810
0.6 10.051 7.569 7.358 7.316 8.873 8.048 10.842 7.413
0.7 10.050 7.453 7.282 7.309 8.831 8.127 10.776 7.421
0.8 10.015 7.467 7.253 7.321 8.998 8.043 10.859 7.481
0.9 9.8671 7.563 7.168 7.350 8.819 7.948 10.737 7.229
1 7.028 4.464 4.267 3.940 5.581 4.679 7.639 4.166
N. Taneja et al. / Int. J. Electron. Commun. (AEÜ) 65 (2011) 338–344 341

Fig. 3. Histogram for original and encrypted image.

• The ciphertext should not be able to correctly decrypt, even if


there is a slight difference in the encryption and decryption key.
i.e. ciphertext should be sensitive to the keys defined/chosen.

Two tests are performed to verify the fulfilment of these basic


conditions of key sensitivity [17]. The first test encrypts the orig-
inal image with a slightly different encryption key, and evaluates
the difference between the obtained encrypted images. The dif-
ference images are evaluated to verify the condition that, “image
encrypted with slightly different keys should be completely differ-
ent”. The second test performs decryption of the encrypted image
with slightly incorrect keys. This aims to subjectively assess the
amount of information leakage from the images decrypted with
the wrong decryption key.
To perform these tests, slightly different keys are generated by Fig. 4. Key sensitivity analysis for encryption process.
keeping the threshold factor constant, and modifying the other
parameters of the original encryption key. In the modified keys,
all parameters, excluding one are kept same as that of the origi-
nal key. These are generated by introducing a variation of ‘1’ in the
integer values and ‘0.1’ in the fractional values.
For the original key K1 : (0.65, 1, 15, 6345921), the modified
keys are expressed as K2 : (0.65, 0.9, 15, 6345921), K3 : (0.65, 1,
14, 6345921) and K4 : (0.65, 1, 15, 6345922). It is to be noted that
as compared to K1 , K2 has a different fractional order, K3 has dif-
ferent Arnold iterations, while K4 has a different seed value. A
different K4 yields a distinct random mask for sign bit encryption.
For convenience, the encrypted image is denoted as Cn , which is
a cipher image obtained using key Kn . Similarly, Dn denotes the
image, decrypted with key Kn .
In the first test, the original image is initially encrypted with
key K1 . The encrypted image C1 for this case is shown in Fig. 4(a).
The original image is then encrypted with slightly modified keys
K2 − K4 . Fig. 4(b)–(d) indicates the corresponding encrypted images
C2 − C4 . It is observed that images encrypted with slightly dif-
Fig. 5. Key sensitivity analysis for decryption process.
ferent keys are completely incomprehensible. Difference images
are then generated to measure the deviation between the image
encrypted with the original key K1 , and the slightly different keys To verify the second condition of key sensitivity, cipher image
K2 − K4 . Fig. 4(e)–(g) illustrates these difference images. Though all C1 is decrypted with key K2 − K4 , instead of the actual key K1 . The
the encrypted images look alike, sufficient deviation exist between decrypted images obtained with these keys are indicated in Fig. 5
them. This verifies that, “ciphertexts generated using slightly dif- (1st three columns). Fig. 5 (last column) illustrates the retrieved
ferent keys are completely different from each other”. image using correct decryption key. The images decrypted with
342 N. Taneja et al. / Int. J. Electron. Commun. (AEÜ) 65 (2011) 338–344

Fig. 7. Decrypted output for noise attacked images.

correct key, it is observed that no details about the original image


could be retrieved from the decrypted output. This illustrates the
sensitivity of the proposed technique towards channel noise.
Analysis reveals that images retrieved in noisy or lossy chan-
nel condition are not even recognizable or of acceptable quality.
Thus, the knowledge of encrypted data and correct decryption keys
Fig. 6. Decrypted output for lossy communication channel. would not be of any use to an intruder, even if certain bits are
not correctly received. This ascertains the strength of the proposed
incorrect keys are completely incomprehensible, and do not leak technique in noisy or lossy communication channel.
any information about the original image. In contrast, decryption
using actual key retrieves the image correctly. This depicts the 4.5. Approximation attack analysis
high key sensitivity of the proposed cryptosystem, and also verifies
the second condition of Kerckhoff’s principle, “decryption using a In an approximation attack, the intruder attempts to reconstruct
wrong decryption key should not reveal any information about the the original data by replacing certain parts of the encrypted data
original image”. with an arbitrarily assumed value [18]. Similar kind of strategy is
The performed test reflects that the proposed cryptosystem followed in the case of lossy communication channel, where the
fulfills the two conditions of Kerckhoff’s key sensitivity, that are lost data is replaced by either ‘0’ or ‘255’. Thus, the performance
essential to ascertain the security of any cryptosystem. of the proposed technique in a lossy communication channel is
treated as analogous to its performance for approximation attacks.
4.4. Performance in lossy and noisy communication channels The approximated data in these cases is the bulk data or the ran-
domly distributed data. The retrieved approximate copy is shown
This section analyzes performance of the proposed technique in in Fig. 6(e)–(h). It is observed that the obtained images do not give
noisy and lossy communication channels. It aims to find the exis- any clue about the original image content and preserve data confi-
tence of avalanche effect in the cipher image, i.e. it evaluates the dentiality. This demonstrates the ability of the proposed technique
sensitivity of the encrypted output towards channel error or noise to withstand an approximation attack.
in two different test scenarios.
In the first test scenario, the communication channel is assumed 4.6. Computational time
as lossy in nature. This causes some part of the encrypted data to
be lost during transmission. Reconstruction is then performed by Different tests are performed to ascertain the benefit of NIE
assigning a value of ‘0’ or ‘255’ to the lost data. Four different data based subband selection, on the security offered or the com-
loss cases are considered. These cases assume a data block of 2 × 2, putational performance. In these tests, the selection criterion is
20 × 20 and 50%, as the lost data blocks. To observe the impact of modified such that encryption is performed on random or all the
the location of lost data on the decrypted output, these blocks are subbands.
considered at different locations. In the first test, instead of NIE based selection, random selection
Fig. 6(a)–(d) indicates the encrypted image with the lost data of subbands is performed. These randomly selected subbands are
blocks, while Fig. 6(e)–(h) illustrates the decrypted output for these then encrypted with the proposed technique. For encryption, the
images. In the case of small data loss, some information about the number of randomly selected subbands is kept same as the num-
original image can be retrieved from the decrypted output, while in ber of significant subbands selected using NIE based criterion. The
the case of large data loss, no details of the image are visible. This encrypted output for different combinations of randomly selected
reflects the robustness of the proposed technique against eaves- subbands is illustrated in Fig. 8 (1st 3 columns). It is observed that
dropping. Even if an intruder has the correct keys and the entire the encrypted output does not provide data confidentiality. Details
encrypted data except some data part, then also, details of the of the original image are clearly visible from the encrypted out-
original image cannot be retrieved. put. Fig. 8 (4th column) indicates the encrypted output, when NIE
In the second test scenario, performance of the proposed tech- based criterion is used to select the subbands for encryption. This
nique is evaluated for a noisy channel. To perform this test, two reflects an incomprehensible image. Despite the same amount of
different cases of noise addition are considered. In the first case, data encrypted for the two cases, data confidentiality is not attained
small amount of random noise is added to the encrypted image. In with a random selection of subbands (Fig. 8). This is owing to the
the second case, Speckle noise, Gaussian noise, or salt and pepper fact that random selection of subbands may not necessarily select
noise is added to the encrypted image. The noise effected encrypted significant subbands in every encryption round.
images are then decrypted, output for which is shown in Fig. 7. To further verify the results, computational time and PSNR
Fig. 7(a) and (b) shows the decrypted ‘Huts’ and ‘Mandril’ image, values are measured for the encryption of randomly selected sub-
for a small noise effected encrypted image. Fig. 7(c) and (d) shows bands. The obtained values are compared with the corresponding
the decrypted image, when speckle noise of variation 0.02 is added values for encryption of NIE based selected subbands. A comparison
to the encrypted image. Though decryption is performed with the is illustrated in Fig. 9. It is observed that there is no significant dif-
N. Taneja et al. / Int. J. Electron. Commun. (AEÜ) 65 (2011) 338–344 343

to small quantum of data being encrypted in the proposed par-


tial encryption technique. It is observed that approximately 25%
data is significant, as per the proposed methodology. As soon as
the amount of encrypted data increases from 25% to 100%, it is
obvious that computational time and resource requirements would
increase.
Further, it is to be noted that though 25% of the data is
significant, the actual data encrypted is far less. In the pro-
posed technique, only sign bits are encrypted. This consumes
only one bit per coefficient. For a 256 × 256 image, encoded at
8 bpp, the amount of data encrypted using the proposed tech-
nique would be (256 × 256 × 1 × 0.25) = 16,384 bits, as compared
to (256 × 256 × 1) = 65,536 bits, if total sign bit encryption is per-
formed and (256 × 256 × 8) = 524,288 bits, if all the pixel values are
encrypted. This indicates that the proposed technique encrypts
only (16,384/524,288), i.e. 3.125% of the entire image data.
On the basis of the above discussion, it is clear that despite the
Fig. 8. Encryption with different selection criterion. encryption of only 3.125% of the entire image data, the proposed
technique ensures perceptually secure image in every encryption
round. This relatively consumes low computational resources for
the security level offered. This also illustrates that the proposed
technique is advantageous over total encryption, as well as the
encryption of randomly selected subbands [11].

4.7. Edge distortion

To prove the attained perceptual security, edge ratio (ER) and


edge deviation ratio (EDR) [14,19] are evaluated. Correct interpre-
tation of these parametric values offers an objective assessment,
analogous to the subjective results of an image-pair. Correspond-
Fig. 9. PSNR and computational time comparison for encryption of randomly and ing to the original image, ER reflects the remaining edges in the
NIE based selected subbands. encrypted output, whereas EDR expresses the deviation in the
edges. These are mathematically represented as follows:
ference in the computational time for the two cases. However, the
N−1
PSNR values with the random selection of subbands is observed 
to be higher, than with the proposed methodology. This reflects B̂(i, j)
that encryption of randomly selected subbands do not maintain ER =
i,j=0
(6)
data confidentiality. This verifies the subjective results illustrated N−1

in Fig. 8. Collectively, this indicates that with the same computa- B(i, j)
tional time, the proposed technique performs better in terms of i,j=0
data confidentiality and PSNR values.
N−1
In the second test, the subband selection is eliminated, i.e. 
instead of only significant subbands, all the subbands are encrypted. |B(i, j) − B̂(i, j)|
Total encryption is performed instead of partial encryption. The i,j=0
EDR = (7)
results are then compared with the results obtained using the N−1

proposed partial encryption technique. For the two techniques, a (B(i, j) + B̂(i, j))
comparative graph for PSNR values and time consumption is indi- i,j=0
cated in Fig. 10. Similar PSNR values are observed for the two
cases. However, computational time consumption shows note- where B(i, j) and B̂(i, j) denote the bit value in the edge detected
worthy changes. It is observed that total encryption consumes binary matrix for the original and the encrypted image, respec-
approximately thrice the computational time, that is consumed tively.
for partial encryption using the proposed technique. This is due The values obtained for ER and EDR evaluation of the proposed
technique is indicated in Table 2. For all the test images, ER (> 0.6)
indicates the remaining edges in an encrypted image. It is observed
that all the test image-pairs yield a high value of EDR (> 0.86). This
reflects that irrespective of the left-out edges, more than 86% of
these edges are displaced from their original location. This further

Table 2
Edge ratio and edge deviation ratio for test images encrypted with key 1.

Image ER EDR Image ER EDR

Crowd 0.904 0.864 Pillars 0.689 0.957


Barbara 0.803 0.943 Butterfly 0.782 0.951
Lena 1.048 0.957 Alonehut 1.006 0.939
Fig. 10. PSNR and computational time comparison for total encryption and pro-
Mandril 0.767 0.949 Peppers 1.039 0.956
posed partial encryption.
344 N. Taneja et al. / Int. J. Electron. Commun. (AEÜ) 65 (2011) 338–344

ascertains and validates the high perceptual security attained by [2] Tang L. Methods for encrypting and decrypting MPEG video data efficiently. In:
the proposed technique. Proc. of the int. multimedia conf., Boston, MA, November. 1996. p. 219–30.
[3] Torrubia A, Mora F. Perceptual cryptography on MPEG layer III bit-streams. IEEE
Trans Consumer Electron 2002;48(4):1046–50.
5. Comparative analysis [4] Lian S, Sun J, Wang Z. A novel image encryption scheme based on JPEG encoding.
In: Proc. of the 8th IEEE int. conf. inform. visualisation, London, UK, July. 2004.
p. 217–20.
This section presents a comparative analysis of the proposed [5] Xin Y, Tao R, Wang Y. Image encryption on a novel reality-preserving fractional
technique with other FRWT and FRWPT based encryption tech- Fourier transform. Proc. of the int. conf. innovation computing, inform. and
niques. Chen and Zhao [10] have suggested the use of fractional control, Beijing, September. 2006. p. 22–5.
[6] Liu Z, Da J, Sun X, Liu S. Triple image encryption scheme in fractional Fourier
order and scaling factor, as the keys to provide complete security. transform domains. Opt Commun 2009;282(4):518–22.
This offers less security and small key space, due to the periodic- [7] Joshi M, Shakher C, Singh K. Logarithms-based RGB image encryption
ity property of fractional order. The limited key space makes the in the fractional Fourier domain: a non-linear approach. Opt Lasers Eng
2009;47(6):721–7.
encryption technique vulnerable to various cryptanalytic attacks.
[8] Mendlovic D, Zalevsky Z, Mas D, Garcya J, Ferreira C. Fractional wavelet trans-
As compared to this, the proposed technique employs fractional form. Appl Opt 1997;36(20):4801–6.
order, Arnold iterations and seed value, as the different components [9] Huang Y, Suter B. The fractional wave packet transform. Multidimensional Syst
Signal Process 1998;9(4):399–402.
of the security key. This yields a large key space. The large key space,
[10] Chen L, Zhao D. Optical image encryption based on fractional wavelet trans-
combined with the high key sensitivity, makes the proposed tech- form. Opt Commun 2005;254:361–7.
nique resistant against various cryptanalytic attacks (demonstrated [11] Chen L, Zhao D. Image encryption with fractional wavelet packet method. Opt
in Section 4). Int J Light Electron Opt 2008;119(6):286–91.
[12] Wickerhauser M. Adapted wavelet analysis from theory to software. A.K. Peters
In another technique introduced by Chen and Zhao [11], Wellesley; 1994.
randomly selected subbands are encrypted to achieve data confi- [13] Tao R, Qi L, Wang Y. Theory and applications of the fractional Fourier transform.
dentiality. However, it is demonstrated in Section 4.6, that random Beijing: THU Press; 2004.
[14] Taneja N, Raman B, Gupta I. Security solutions for still visual data. Ph.D. Thesis.
selection of subbands is not an appropriate approach. The randomly Indian Institute of Technology Roorkee, India.
selected subbands may not always lead to selection of significant [15] G. Peterson, Arnold’s cat map, 1997. Available from:
subbands. Hence, this does not ensure perceptual security in all http:online.redwoods.cc.ca.us/instruct/darnold/maw/catmap3.htm.
[16] Marion A. An introduction to image processing. London: Chapman and Hall;
encryption rounds (Fig. 8: 1st–3rd columns). The proposed tech- 1991.
nique, however, selects subbands on the basis of their NIE content. [17] Schneier B. Applied cryptography, second edition: protocols, algorithms and
Encryption of selected significant subbands yields perceptually source code in C. New York: Wiley Publications; 1996.
[18] Mao Y, Wu M. A joint signal processing and cryptographic approach to multi-
secure output in every encryption cycle (Fig. 8: 4th column).
media encryption. IEEE Trans Image Process 2006;15(7):2061–75.
Further, the technique introduced in [11] exhibits low key [19] Taneja N, Raman B, Gupta I. Chaos based partial encryption of SPIHT compressed
sensitivity. Details of the original image are visible, even when images. Int J Wavelets, Multiresolution and Information Processing; in press.
decryption is performed using an incorrect key [11]. However, the
proposed technique overcomes this weakness, and demonstrates Nidhi Taneja received her B.E. degree in Electronics &
Communication and M.Tech degree in Digital Communi-
that decryption with a slightly different security key does not reveal cation in 2001 and 2006, respectively. At present, she is
any information of the original image (Fig. 5: 1st–3rd columns). pursuing her Ph.D. in Department of Electrical Engineering
This proves an improvement by the proposed technique, over the at Indian Institute of Technology Roorkee, India. Her area
of interest includes Wireless Communication, Multime-
existing fractional combined domain based encryption techniques. dia Transmission over Packet Networks, Image Encryption
and Visual Cryptography.
6. Conclusion

This article presents a FRWT domain based technique. This


considers the inherent properties of image data and selectively Balasubramanian Raman received his Ph.D. in Math-
encrypts only the significant part. To identify the significant data, ematics (2001) from Indian Institute of Technology,
relationship between NIE and perceptual information of a subband Madras, India. At Present, he is an Assistant Professor
and Head of the Computer Vision, Graphics and Image
is derived. An increased effective key space is obtained by chaotic Processing Laboratory in the Department of Mathemat-
encryption of the selected subbands. This makes the fractional ics at Indian Institute of Technology Roorkee, India. He
order, just a part of the key, but not the most important key com- worked as a Post Doctoral Associate in ECE Department,
and member of the Visualization Research Laboratory (VIZ
ponent. With the proposed technique, encryption of only 3.125% Lab), at Rutgers, The New State University. He was also a
of the entire image data leads to an acceptable perceptual secu- Post Doctoral fellow of Computer Engineering and Com-
rity. The proposed technique provides high cryptographic security, puter Science (CECS), and member of the Computational
Intelligence Research Laboratory (CIRL), at the University
that is ascertained by considering the statistical and key-related of Missouri-Columbia (MU), Missouri, USA. He has also
attacks. Performance analysis in lossy and noisy communication worked as Visiting Professor in Department of Electrical and Computer Engineering
channels demonstrate its robustness against eavesdropping and at University of Windsor, Canada under Boyscast Fellowship. His areas of research
includes Computer Vision, Graphics, Satellite Image Analysis, Scientific Visualiza-
approximation attacks. tion, Imaging Geometry, Reconstruction Problems, Image Encryption and Digital
Watermarking.
Acknowledgement
Indra Gupta received her B.Tech degree in Electrical
The first author acknowledge the financial support provided by Engineering from HBTI, Kanpur, in 1984. She com-
pleted her M.E. and Ph.D. from University of Roorkee,
Ministry of Human Resource Development, Government of India. India. She is Currently an Associate Professor in the
Department of Electrical Engineering, Indian Institute
References of Technology Roorkee, India. Her areas of interest
includes Advanced Microprocessor Applications, Infor-
mation Security, Multimedia Processing, Process Control
[1] Van Droogenbroeck M. Partial encryption of images for real-time applications. Applications, Biomedical Imaging, Content based Image
In: Proc. of the 4th benelux signal process. symp., Hilvarenbeek, The Nether- Retrieval and Online Computer Applications.
lands. 2004. p. 11–5.

You might also like