16th international conference on Sciences and Techniques of Automatic control STA'2015-PID3815-BSP

& computer engineering - STA'2015, Monastir, Tunisia, December 21-23, 2015

Hybrid SVD- DWT watermarking technique using AES algorithm

for medical image safe transfer

Sondes AJILI, Mohamed Ali HAJJAJI, and Abdellatif MTIBAA



Abstract— This manuscript presents a watermarking method The basic principal of watermarking method, applied on
based on Discrete Wavelet Transform (DWT) and Singular medical domain, consists to integrate the medical data
Value Decomposition (SVD) with the Advanced Encryption (private patient information) specific to the patient into the
Standard (AES) algorithm encryption applied on medical host medical image. In fact, the difficulty of the watermarking
image. The proposed schema will be developed to increase in the medical domain is preserving the quality of the image
the protection and robustness along with security. In our together with the extracting of the patient’s information (the
algorithm, after decomposing the original image into four totality of data) after applying several attacks. In order to
sub-bands in the second level of the DWT, we apply the SVD increase the security of medical images the encryption
transform to the high band, and embed the binary watermark techniques are used with watermarking method. Since, the
(patient’s data) by changing the singular values. For enhance encryption algorithms are considered as a complementary
robustness, AES encryption algorithm are used. After protection technique. In this context, we developed a security
application various type of attacks (JPEG compression, method based on watermarking and AES encryption
median filter, Salt and pepper, and cropping), we used the techniques for medical images using SVD and DWT. Our
serial turbo code in order to control identification and, if goal is to join the robustness of watermarking technique with
possible, correct the watermark, the security offered by the encryption algorithm.
Experiments and analysis results demonstrated that low The rest of this manuscript is planned as follows. In Section 2,
distortion and increased robustness have been made in the a review of related works concerning the use of SVD and
process of watermarking, simultaneously, elevated security DWT in watermarking schemes with encryption techniques
has been proved in the encryption phase. The evaluation together is given. The performance measures is introduced in
studies are applied on MRI medical image. section 3, whereas, the description algorithm is described in
detail in Section 4. Section 5 discusses the results obtained
Keywords—DWT; SVD; AES; Medical image; security; using our proposed method. Finally in Section 6, the
attacks; serial turbo code concluding remarks are given.
I. INTRODUCTION II. A REVIEW OF RELATED WORKS
Recent advances in digital technology have allowed In this part, We describes the different works in the
transmission and dissemination of digital services such as literatures which had been carried out on the combination of
(Military, Medical...) efficiently and easily to distant places. watermarking method with encryption algorithm by applying
Indeed, a majority of medical images today exist in electronic the Singular Value Decomposition and Discrete Wavelet
format for convenient storage and retrieval. Computer Transform in the watermarking schema and Advanced
networks make it highly practical to access and share Encryption Standard algorithm in the encryption technique.
information between medical personnel, lead to improved in In [3] Murty proposed A robust image watermarking method
the quality of patient care. However, the ease of such using combination of SVD, DWT and DCT Technique. HL
manipulations increases some security issues (availability, frequency band of the image was selected for the embedding
confidentiality and reliability). In confidentiality issue, the process.
information of the patient is accessible to only authorized Kumar et al. in [4] have proposed a watermarking method
users. The reliability issue is founded on authenticity and based on SVD and DWT. In this proposed method, three level
integrity. The availability issue assures the medical data must of DWT have used. After that, the watermark is embedded
be accessed. This provides a demonstration that the medical into the modified singular values of original image. Medium
data has not been changed or hacked by the illegal users. To frequency bands (HL3, LH3) were chosen for embedding.
address this security issue, several of techniques for the Zhou and Jin [2] proposed a zero-watermarking and
protection of rightful ownership of medical images have been SVD. The characteristic are extracted from the host image by
proposed among them watermarking [8], [7], [9] and using DWT and SVD to each block. The embedding of this
encryption [10], [11], [12]. method consists in using the exclusive among the pixel of the
binary watermark, and singular value of each block
Sondes AJILI, Electronics and Microelectronics Laboratory, University sequentially.
of Monastir, Tunisia ([email protected]) In [5], Makbol et al. present a novel robust blind
Mohamed Ali HAJJAJI, Higher Institute of Applied Science and
Technology of Kasserine, University of Kairouan
watermarking method based on the Redundant DWT and
([email protected]) SVD. The image watermark was integrated in the matrix S of
Abdellatif MTIBAA, Electronics and Microelectronics Laboratory, the redundant DWT subbands of the original image.
University of Monastir, Tunisia ([email protected])

978-1-4673-9234-1/15/$31.00 ©2015 IEEE 69

Authors in [1] proposed a robust watermarking schema using (Normalized Cross-Correlation) [14]. Its value is generally
combination of Singular Value Decomposition, Discrete between 0 and 1. Ideally it should be 1.
Wavelet Transform and Discrete Cosine Transform. The first ¦¦ W ij  ww W ij  ww (3)
step consists to decompose the original image into 1st level NCC W ,W i j

DWT monitored by transformation of LL (Low frequency § ·§ ·

¨ ¦¦ Wij  ww ¸¨ ¦¦ W ij  ww
2 2

band). The second step is to watermarking the transformed ¸

© i j ¹© i j ¹
image by using SVD and DCT. After that the singular value
of watermarked image is integrated in the singular value of
the original image. To obtain the watermarked image, it w
w w
Where w , w , w , correspond, respectively, the original
watermark, the extracted watermark, the mean of the original
suffices to use the inverse SVD on original V, U vectors and
modified S vector followed by inverse DWT and inverse watermark and the mean of the extracted watermark.
DCT. The performance of the encryption method is evaluated
based on it robustness against various attacks by calculating
Recently, Ray et al. [6] proposed a new algorithm based
the information entropy and the histogram.
on SVD and DWT technique for image watermarking
schemes. The first step consists to decompose the original D. The entropy
image into sub-bands by using the discrete wavelet transform. Entropy is a magnitude characterizing the quantity of data
Then, the watermark image is integrated in all of sub-bands contained in an image. In fact, the information entropy is the
by changing the S vector of each sub-band. The third step is most important feature of randomness. In fact, if the
to encrypt and embed the S vector modified of the watermark dispensing of grey values is very uniform, the information
image in place of original S vector. For the encryption entropy is greater [15]. The entropy E(s) is defined as
process, RSA algorithm as been used. follows:
N
III. PERFORMANCE MEASURES E(s) ¦ ª¬ P(Si) log2 P Si º¼ Bits (4)
i 1
The performance of the watermarking method is
evaluated based on its imperceptibility robustness. E. Histogram
A. PSNR (Peak Signal to Noise Ratio) A histogram is a statistical curve showing the distribution
of pixels according to their value. The histogram is very
A PSNR indicates that the original and the watermarked
helpful to control the exposure of an image.
image are more similar. In fact, this resemblance means that
the message inserted is more imperceptible. In general, if the IV. PROPOSED COMBINED DWT-SVD WATERMARKING
PSNR value of watermarked image is greater or equal to 27, ALGORITHM USING AES ENCRYPTION
then the PSNR is acceptable [13]. The PSNR is defined as
In this sector, as shown in figure1, we explain our
follows:
proposed schema of combined Discrete Wavelet Transform
max I 2 (i, j ) (1)
( PSNR) dB 10log10{N u M [ ]} with Singular Value Decomposition used for medical image
¦ [ I (i
i, j
, j )  I w (i , j )]2
watermarking by embedding the totality of the patient’s data
(watermark).

B. WPSNR (Weighted peak signal to noise ratio) Original 2-D SVD Insertion SVD-1 2-D AES
image DWT Step DWT-1 Encryption
The WPSNR is founded on the comparison between the
watermarking image and the host image (pixel to pixel). Watermark Treatment
Watermarked and
encrypted Image
WPSNR is defined as.
M u N max I 2 (i, j ) Diffusion
(WPSNR)dB 10log10{ } (2)
I (i, j )  I w (i, j ) 2
¦i, j
[
1  varI (i, j)
] Extracted
patient’s
information
Extraction
Step
Watermarked
and attacked
Image
AES
Deception

With: I(i,j) representing the intensity value of the pixel(i,j)

from the host image, var(i,j) is the variance of pixel (i,j) and
Iw (i,j) is the intensity value of the pixel of the testing Figure 1. Combined DWT-SVD watermark embedding algorithm
image. N and M are respectively the width and height of the
host image. Before the insertion step, the information of patient
undergoes several phases with the goal is increasing the
C. NCC (Normalized Cross-Correlation) degree of embedding and ensuring the extraction of the
The robustness of the watermarking system depends of watermark after applying much type of attacks. The proposed
correlation factor. Indeed, the resemblance between the method is divided into two phases:
original and extracted watermark is detected by the calculus
of different distances among the detected and inserted
watermark. This measure is performed via the NCC

70
A. Combined DWT-SVD Watermark Embedding Process x Dividing the sub-band LH2 into sub-blocks each sub-
Algorithm block is formed of 4x4 pixels
In view to obtain a best imperceptibility, it must pass with x Dividing the sub-band LH2 into sub-blocks each sub-
two steps. In fact, the first step consists in preparation of the block is formed of 4x4 pixels.
information for inserting in medical image; the patient’s
information such as: file number, name, diagnosis, etc. and x On each sub-block, a vector decomposition is
the hospital center’s name. The second step consists to applied, the Singular Value Decomposition (SVD), to
preparation of the insertion space; the pretreated of the obtain the matrix S.
original image to lastly achieve the embedding of watermark.
x We choose the second coefficient S "σ2" to support
The preparation of information inserting: The diverse steps
the Watermark, which aims to achieve a maximum
of the preparation of information inserting, as given in figure
compromise between robustness and invisibility.
2, are realized as follows.
x Generate a binary signature sized of 160 pixels x Each watermarked matrix S will be used again for the
detained by the hospital center; our goal is to control reconstitution of the blocks by the inverse SVD.
the authenticity of patient's file. The SHA-1 (Secure
Hash Algorithm) is applied [16]. x Watermarked blocks will be put back to form
watermarked LH2.
x Convert the rest of the patient’s information into a
binary sequence. x Application of Discrete Wavelet Transform to form
in the end the watermarked image.
x The binary sequence and the binary signature are
concatenated to obtain a binary message (the Original
Image
watermark).this message will be inserted thereafter in Decompositio
the transformation domain of the image. LL2 LH2 LH2 n in 4x4 block
size
2-D
DWT
x Coding information message (watermark) via the Second level
HL2 HH2
ECC (error correcting codes); in our approach we
apply the serial turbocode [17] whose principal goal
is to exact the totality of our watermark after
application of several attacks.
Singular value decomposition for
each block

Patient’s Hospital U S V
informatio center

ªV1 0 0 0 º
Binary SHA-1 «0 V 0 0»
data Function « 2 »
« 0 0 V3 0 »
« »
¬ 0 0 0 V4 ¼
Binary
signature
Figure 3. The preparation of the insertion space
Concatination
The Embedding Process Algorithm: The several steps of
embedding process algorithm, as mentioned in figure 3(b), are
Codage
ECC described as follows.
x Adding to the S vector the information related to the
Coded patient.
massage

V ic V i  D *Wi (5)
Figure 2. preparation of information inserting
th
With: σi: being the i old coefficient to support the
The preparation of the insertion space: The different steps watermark; σ’i: being the ith new watermarked
of the preparation of insertion space, figure 3 (a), are coefficient; Wi: being the ith bit to hide; α: being
performed as follows. named often the visibility factor. Indeed, α is an
x Applying the technique of the Discrete Wavelet important factor in the watermarking system. If this
Transform and divide the host medical image into factor is big, we win in terms of the robustness
four subbands i.e. LL2, LH2, HL2, and HH2.The however we lose in terms of imperceptibility and vice
second level of decomposition Discrete Wavelet was versa.
chosen.
x Each watermarked matrix S will be used again for the
x For better extraction, we choose the sub-band LH2 to reconstitution of the blocks by the inverse SVD.
support the Watermark.

71
 Watermarked AES Decrypted 2-D DWT
x Watermarked blocks will be put back to form and encrypted
image
algorithlm Watermarked
image Second level LL2 LH2watermarked

watermarked LH2.
Key LH2watermarked
HL2 HH2
x Application of Discrete Wavelet Transform to form
the watermarked image. Decomposition
in 4x4 block
x The watermarked image is finally encrypted by using size
Singular value decomposition for
AES (Advanced Encryption Standard ) [18]. each block

Singular Value Decomposition

Patient’s U S V
Watermarked Data
and encrypted U S V
image

W
Factor of If : moy  V 2 Calcul of the average:
visibility α Decoded Extracted
Key AES
Insertion step message watermark Bit=0 V1  V 3
algorithlm
σ2-watermarked= σ2+αW(i)
moy
Verification else Bit=1 2
ith bit of W
and
Watermarked
image ªV1 0 0 0 Decision
«
« 0 V 2 _ watermarked 0 0
«0 0 V3 0

Inverse 2-D
«
¬0 0 0 V4
Figure 5. Extraction Process Algorithm
DWT
S’watermarked
V. EXPERIMENTS RESULTS AND DISCUSES
LL2 LH2watermarked Singular Value Decomposition
in Reverse
ith wtarmarked block
A. Experiments results of the watermarking algorithm
HL2 HH2 sized 4×4

LH2watermarked
It is extremely essential to obtain good values of PSNR,
WPSNR and NCC regardless of the domain or the method for
dissimulated of the watermark. Indeed, this is particularly true
in the medical imaging domain. For the validation of our
Figure 4. The Embedding Process
proposed algorithm, we have been used the IRM image
(512×512 pixels). This work permit to detect the whole of the
B. Combined DWT-SVD Watermark Extraction Process watermark inserted in the tested medical image. In the
Algorithm literature, we obtain an acceptable extracted watermark if the
value of the NCC is greater or equal than “0.75” [19].
After a stage of diffusion in which on the watermarked
Contrariwise, in the domain of medical image, the watermark
encrypted image can suffer several types of attacks, an
is a binary message. In fact, in the extraction phase, it is
extraction phase of the insertion data from the watermarked
obligatory to obtain the totality of watermark inserted (NC=
and encrypted image will take place (figure 5). In general, the
1).Otherwise, we get a false diagnosis. The measuring the
extraction stage follows the reverse steps with respect to the
PSNR and the WPSNR are acceptable when the rate of PSNR
insertion. To ensure more robustness of our algorithm, we
and WPSNR gives an value exceed 35dB [20].We begin
chose to use blind extraction i.e. without resorting to the host
analyzing the experimental results by calculate the NCC,
image to extract the message inserted.
PSNR and WPSNR, as shown in figure 6, without the
x Decrypt the watermarked encrypted image by using
application of attacks on the watermarked images.
AES to obtain the watermarked image.
x Decomposing the watermarked image IW with using
Discrete Wavelet Transform to extract the
watermarked sub-band; LH2Watermarked.
x Dividing the sub-band LH2Watermarked into sub-blocks
each sub-block is formed of 4x4 pixels.
x Applying SVD on each block LH2Watermarked. (a) (b) (c)
PSNR = 57,6544; WPSNR = 59.799; NCC =1
x Calculate the average between the first ;(σ1) and the
Figure 6. Without attacks: Rate imperceptibility and Robustness of the
third (σ3) value of the component of SWatermarked. watermark; (a) original IRM image, (b) encrypted watermarked image, (c)
decrypted watermarked image.
x If the average is lower than σ2, than the detected
watermark bit is '0'. Else the detected watermark bit is The watermark, in the domain of medical image, should
‘1’. be imperceptible. In fact, the watermarked image and the host
x Extraction the binary message and applying the serial image must be similar. Our algorithm will be more significant
Turbocode. whether it will be tested and validated against several types of
attack. After applying of each attack, we extract the
x The patient data is extracted, verified, and eventually embadded watermark and compared with the original
corrected. watermark. The aim is to ensure that watermark is not
modified by the attacks applied on the watermarked image.
Consequently, regardless the system of watermarking, it is

72
extremely important to get a excellent value of PSNR, 1

WPSNR and an NCC=1. 0,9

Among the most important watermarking constraints is

NCC
0,8

the Robustness against several types of attacks. We have 0,7

chosen to apply the most common attacks to properly analysis 0,6

the robustness of our system which are: JPEG attacks, [16*16] [32*32] [64*64] [128*128] [256*256]
Window size
Gaussian noise, and Cropping.
Now, after attacking the MRI medical image, we attempt Figure 11. Evaluatiion of the NCC values of the encrypted, watermarked
and attacked MRI image by cropping
to extract our watermark and calculate the value of the NCC.
The goal of this calculation is to determine the degree of
robustness of our algorithm to resist to several attacks. 1

Figures 7, 8, 9, 10, 11, 12, 13 and 14 present the simulations 0,9

results of the WPSNR, PSNR and NCC values between the

NCC
0,8

extracted and original watermark after attacks. 0,7

60
PSNR
WPSNR 0,6
50
0,5
[3 3] [4 4] [5 5] [6 6] [7 7] [8 8] [9 9]
PSNR and WPSNR

40 Filter’s window size

30
Figure 12. Evaluatiion of the NCC values of the encrypted, watermarked
20
and attacked MRI image by median filter
10

0
0 [16*16] [32*32] [64*64] [128*128] [256*256]
1
Window size

0,95

Figure 7. Evaluatiion of the WPSNR and PSNR values of the encrypted, 0,9
NCC

watermarked and attacked MRI image by cropping 0,85

0,8
40
PSNR 0,75
WPSNR
35
0,7
30 1 2 3 4 5 6 7 8 9
Ratio (%)
PSNR and WPSNR

25

20
Figure 13. Evaluatiion of the NCC values of the encrypted, watermarked
15
and attacked MRI image by a JPEG compression algorithm.
10

0 1
0 [3 3] [4 4] [5 5] [6 6] [7 7] [8 8] [9 9]
Filter’s window size

0,9

Figure 8. Evaluatiion of the WPSNR and PSNR values of the encrypted,

NCC

0,8
watermarked and attacked MRI image by median filter
0,7
45
P SNR
WP SNR 0,6
40

35 0,5
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
PSNR and WPSNR

Density
30

25

20 Figure 14. Evaluatiion of the NCC values of the encrypted, watermarked

15 and attacked MRI image by Salt and pepper noises
10

0
B. Experiments results of the encryption algorithm
0 10 20 30 40 50 60 70 80 90 100
Ratio (%)
The IRM image encrypted and transformed into an
Figure 9. Evaluatiion of the WPSNR and PSNR values of the encrypted, illegible 2D sequence as showed on figure (15). The
watermarked and attacked MRI image by a JPEG compression algorithm
histogram of the encrypted and the original image are also
40
PSNR illustrated.
WPSNR
35
2500

30 10000
PSNR and WPSNR

25 2000

8000

20
1500

6000
15
1000

10 4000

5 2000
500

0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0
0

Density
0 50 100 150 200 250
0 50 100 150 200 250

Figure 10. Evaluatiion of the WPSNR and PSNR values of the encrypted, (a) (b) (c) (d)

watermarked and attacked MRI image by Salt and papper noises

Figure 15. (a) IRM Image, (b) Histogram of the original image, (c) The
encrypted Image, (d) Histogram of the encrypted image

73
 As the results exposed in figures 15, the histogram of the References
encrypted watermarked image have a regular intensities [1] Singh, Amit Kumar, Mayank Dave, and Anand Mohan. "Hybrid
distribution. In fact, Comparing the histogram of the Technique for Robust and Imperceptible Image Watermarking in
DWT–DCT–SVD Domain."National Academy Science Letters 37.4,
watermarked encrypted image with the histogram of the host pp. 351-358, 2014.
image we find that the probabilities of occurrence of every [2] Zhou, Yaxun, and Wei Jin. "A novel image zero-watermarking scheme
gray level are uniformly repartees. This demonstrates the based on DWT-SVD." Multimedia Technology (ICMT), 2011
effectiveness of our proposed method that equalizes the International Conference on.IEEE, pp. 2873-2876, 2011.
[3] S. Murty, Dr. Rajesh Kumar, “A Robust Digital Image
statistical distribution of the image in a random way making it WatermarkingScheme using Hybrid DWT-DCT-SVD Technique”,
indecipherable. Then, the entropy value is high and equal to IJCSNS,Vol.10,No.10, pp. 185-192, Oct 2010.
7.992. [4] Satendra Kumar, AshwiniSaini, Papendra Kumar, “ SVD based
RobustDigital Image Watermarking using Discrete Wavelet
C. Discussion of the proposed method Transform”, IJCA,Vol. 45No. 10, pp.7-11, May 2012.
[5] Makbol, N.M., Khoo, B.E.: Robust blind image watermarking scheme
In this section, the proposed system is studied by comparing based on redundantdiscrete wavelet transform and singular value
our results to those cited in the section of a review of related decomposition. Int. J. Electron. Commun.(AEU) 67(2), pp. 102–112,
works. Comparing the psycho-visual quality of the original 2013.
and the watermarked medical image, the proposed algorithm [6] Ray, Arun Kumar, et al. "Development of a new algorithm based on
svd for image watermarking." Computational Vision and
provides very good results. In the absence of any type of Robotics.Springer India, pp.79-87, 2015.
attack; the PSNR equal to 57,6544, WPSNR equal to 59.799 [7] Navas KA, Sasikumar M, “Survey of medical image watermarking
for IRM are obtained, which gives results that are often better algorithms”, 4th InternationalConference: Sciences of Electronic,
than those algorithms previously shown in the related section Technologies of Information and Telecommunications, pp. 1–6,2007,
TUNISIA.
of the review works. [8] C. I. Podilchuk and E. J. Delp, “Digital Watermarking:Algorithms and
After applying several attacks, it is necessary to evaluate the Applications,” IEEE Signal Processing Magazine, July 2001, pp. 33-
NCC. We are starting testing our algorithm against the 46.
median filter. The test is evaluated with various sized [9] I. Cox, M. Miller, and J. Bloom. "Digital Watermarking: Principles &
Practices". Morgan Kaufmann Publisher, San Francisco, USA, 2002.
windows (from [3 × 3] until [9 × 9]). The good obtained [10] Borie, J. C., Puech, W., & Dumas, "Crypto-Compression System for
results of the NC (equal to 1) when the sized windows go Secure Tranfer of Medical Images", In MEDSIP'04: 2nd Medical
from [3 × 3] to [8 ×8].Among the most serious attacks we Image and Signal Processing, pp. 327-331, 2004.
applythe JPEG attack. The result is very effective against this [11] X. Zhang, “Reversible data hiding in encrypted image,” IEEESignal
Processing Letters, vol. 18, no. 4, pp. 255–258, 2011.
kind of attacks. In fact, The NCC value stills equal to 1 when [12] W. Hong, T.-S.Chen, and H.-Y. Wu, “An improved reversibledata
the rate of image compression goes from 10% to 70%. To hiding in encrypted images using side match,” IEEE Signal Processing
evaluate our method against the Salt and pepper noises, we Letters, vol. 19, no. 4, pp. 199–202, 2012.
obtained very promising results; the NC =1 when the density [13] Singh AK, Dave M, Mohan A, “Hybrid technique for robust and
imperceptible image watermarking inDWT- DCT-SVD domain”,
goes from 0.01 to 0.07. We also tested our method against NatlAcadSciLett 37(4), pp. 351–358, 2014.
cropping. The obtained NC values is equal to 1. [14] Kannammal, A., and S. Subha Rani. "Authentication of medical
images using integer wavelet transforms." International Journal of
VI. CONCLUSION Emerging Technology and Advanced Engineering 2.9, pp. 104-108,
2012.
The integrity, confidentiality and privacy of medical [15] Habiba LoukilHadjKacem and Mohamed Salim Bouhlel. Mesure de la
image constitute key factors to consider in the dispensing of qualité des images par l'utilisation de la loi de weber. In International
medical image. We have proposed, in this manuscript, a Conference: Sciences of Electronic, Technologies of Information and
combining SVD and DWT schema for integrating patient's Telecommunications, 2005.
[16] Federal Information Processing Standards Publication, “Secure Hash
data in medical images. The first stage consisted to insert a Standard,” 180-2, 2002.
delicate watermark into the original medical image; the [17] S. Gastan, Codage de canal pour les communications optiques [M.S.
objective is to verify the traceability and the integrity of the thesis], 2009.
medical image.For enhancing the robustness of our method [18] J. Deamon and V. Rijmen, “The Design of Rijndael. AES - The
Advanced Encryption Standard, Springer - Verlag, 2002.
and preserve the confidentiality of patient’s information, the [19] M. J. Mateo, D. J. O’Callaghan, and C. P. O’Donnell, “Comparison
AES encryption algorithm is used. The experiments results between red-green-blue imaging and visible-near infrared rebectance
demonstrate that our proposed schema has a better robustness as potential process analytical tools for monitoring syneresis,” Journal
against several types of attacks. This robustness is achieved of Dairy Science, vol. 93, no. 5, pp. 1882–1889, 2010.
[20] K. Amal and S. Hamad, “A robust non-blind algorithm for
through : the serial turbocode to enhance the security of the watermarking color images using multi-resolution wavelet
inserted message (watermark) against attacks, the hash decomposition,” International Journal of ComputerApplications, vol.
function SHA-1, the Discrete Wavelet Transform space and 37, no. 8, pp. 33–39, 2012.
theSingular Value Decomposition to improve the watermark
integrity. This algorithm is completely adapted to the domain
of medical image since it benefits the utilize of several
algorithms with the aims of ensuring the extraction of all
patients’ information and preserves the quality of the image.

74