Hybrid Transforms Based Watermarking, Encryption and Compression of Bio-Medical Images

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 09 | Sep 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 1333
Hybrid Transforms Based Watermarking, Encryption and Compression
of Bio-Medical Images
T. Spandana1, R. Gurunadha2
1M. Tech scholar, Dept. Of Systems and Signal Processing, JNTU-GV College of Engineering Vizianagaram(A),
Andhra Pradesh, India.
2Associate Professor, Dept. of Electronics and Communication Engineering, JNTU-GV College of Engineering
Vizianagaram (A), Andhra Pradesh, India.
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract –In this paper, we arediscussingabouttheHybrid
transforms-based watermarking, encryptionandcompression
of bio-medical images. The entire system is designed using
MATLAB Software with the version of 2021a. Nowadays,
Medical information plays an important role in medical
diagnosis, transferring of that information should be done in
an encrypted manner. Proposed novel Image Watermarking
which is based on the Discrete Wavelet Transform (DWT),
Hessenberg Decomposition (HD) and Singular Value
Decomposition (SVD). Inverse Discrete Wavelet Transform
(IDWT) used for retrieval of image. In this, watermark is
embedded intosub-bandscoefficient. Sub-bandcoefficientsare
marked by adding a watermark of the same size of the three
sub-bands. Comparison of embedding a watermarkatvertical
(LH), horizontal (HL) and diagonal (HH) details. Here,
Compression scheme applied on the watermarked image to
reduce size of the data without losing its quality. AES
encryption is used for security and proposed methodology
analyzed on data sets of MRI of brain images and Retina
images. The performance metric is analyzed after de-
watermarking with different ordersof Daubechieswaveletsby
using parameters like Peak Signal to Noise Ratio (PSNR),
Mean Square Error (MSE), Root Mean Square Error (RMSE)
and Compression Ratio (CR).
Key Words: Image Watermarking,DWT,HD,SVD,IDWT,
Compression, AES Encryption
1.INTRODUCTION
Watermarking is a process of embedding
information into a digital signal such as image, video,
audio data to easily identify the ownership of the original
data. Therefore, it will be embedded in a way which makesit
difficult to be visualize with the human eye and to be
removed. As computers are more and more integrated via
the network, the distribution of digital data is becoming
faster, easier, and less effort to make exact copies. One ofthe
present research areas is to protect digital watermark inside
the information so that ownership of information will not be
claimed by the third party. The working domain of the
watermarking is either spatial or frequency. Medical data
exchange between the departments is not just hospitals, but
between hospitals in different geographical areas. Medical
images require strict security measures. The medical
information can often be shared to the professionals to
improve treatment. Huge amount medical data should be
processed in hospitals for clinical and medical purpose of
research. Malicious attacks on this medical data repots
should be avoided and also can create medical images with
watermarks certification. Image quality is distorted in the
spatial and transform domains. The process of insertingand
extracting watermarking process. In the original image, a
watermark (hidden image information)isinserted.IfImages
are retrieved from the database, allowing physicians to see
Authentication Phase with Extraction watermark. However,
upon successful watermark extraction, created and the
doctor can proceed with the diagnosis trust.Developmentof
the DWT-HD-SVD Algorithm Based on the watermark used
for this study. A series of biomedical images are hidden
under the cover image. Embedding and retrieval of
watermarks in data can be analyzed. This transformation
includes frequency and Location information, as opposed to
the Fourier transform. Here, Signal energyisconcentratedin
specific wavelet coefficients with the discrete wavelet
transform. This property is useful when image compression
is needed. lossless compression is used to Minimize data
size. Encryption and decryption are Used the security.
Proposal performance of medical data storage is also
evaluated by multiple metrics such as mean squared error
(MSE), peak signal signal-to-noise ratio (PSNR), root-mean-
square error (RMSE), and Compression ratio (CR).
2. LITERATURE SURVEY
A safer and more robust digital watermarking
method based on encryption algorithms and asymmetric
RSA Encryption techniques have been proposed to protect
hidden things (DWT) and singular value decomposition
(SVD). The power to estimate embeddinga watermark in the
low-frequency sub bands of the host image. Transmitted
biomedical images and visual data are Watermark
encryption algorithm and use of thisalgorithmcompression,
watermarking, Encryption to protect data. medical imaging
combined with security analysis of A study of
encryption/watermarking (E/W) systems found that
Wavelet transform (DWT) and singular values
Decomposition Techniques to Reduce Ambiguity (SVD)

Guaranteed by singular value decomposition (SVD). Digital
Watermarking Method Combining Discrete Wavelets
Transformation by Singular Value Decomposition (DWT)
Show that DWT-SVD produces good images uses block
encryption techniques to achieve this. Watermark image
encryption and compression, the method relies on
watermarks. Distributors need to know each user`s
encryption key to complete the process processing,
management and distributionofprivatekeys;Errortryingto
improve watermark security Watermark extraction takes
time. proposed a new image watermarking approach based
on DWT-HD-SVD, AES encryption are trying to improve
watermark security and watermark extraction.
3. PRELIMINARIES
This section presentsthethreetechniquesDWT,HD,and
SVD used in the proposed watermark method. Multi-
resolution timescale signal of DWT can improve watermark
performance. HD performs as the matrix transforms have
further improved robustness. In addition, SVD-based
watermarking schemes have improved performance in
defending against geometric attacks.
3.1 Discrete Wavelet Transform
The Discrete WaveletTransformisthetransformusedin
both Numerical and functional analysis. Wavelet is sampled.
This transform uses discrete values. This conversion
includes Advantages of collecting both frequency and
location information, Contrast with the Fourier transform.
signal energy Focus on specific wavelet coefficients in
discrete Wavelet transform (DWT). This property is useful
for images compression. A multi-resolution representation
called DWT used to Decode step by step from low resolution
to high resolution. DWT separates the signal intotwohalves.
High Frequency and low frequency. high frequency segment
represents the edge component,andthelowfrequencyis the
section and is again divided into a high frequency section
and a low frequency section. Since the human eye is not
sensitive to edge variations, High frequencycomponentsare
usually used.
Fig -1: DWT Sub-bands
Here, LL band filter in 1-stage DWT decomposition
contains a lot of information from the original image.
Vertical, horizontal, skew(diagonal) information of the
original images is included in the LH, HL, and HH bands. LL
band Only images can reproduce the original image. Other
bands are ignored.
3.2Hessenberg Decomposition
HD is a type of matrix decompositions can be used for
square matrix decomposition. A n × n square matrixXcan be
decomposed by using HD as shown by
P H PT = HD(X),
where P is an orthogonal matrix and H is an upper
Hessenberg matrix, and hi,j = 0 when i > j + 1. HDiscomputed
by the Householder matrices. Householder matrix Q is an
orthogonal matrix and it is expressed as
Q = (In − 2µµT)/µT µ,
where µ is a non-zero vector in Rn, and I n is a n × n
identity matrix. n−2 steps are in the overall procedure.
Therefore, HD is computed as
P = (Q1Q2 . . . Qn−2)T X (Q1Q2 . . . Qn−2)
⇒ H = PT X P
⇒ X = P H PT
3.3 Singular Value Decomposition
It is a factorization of that matrix into three matrices in
linear algebra. It possesses some unique algebraic
characteristics and communicates key geometrical and
theoretical insights into linear transforms. It is also having
several interesting data science applications. It is a signal
processing technique that is widely employed. SVD is used
for noise reduction and image compression, among their
things.
A=U ∑ V
A is a m x n matrix that you obtained from an image or
another data source. The orthogonal matrices U and V are
orthogonal matrices and ∑ is a diagonal matrix. Finding the
eigenvalues and eigen vectors of A and is the first
step in calculating the SVD.
The columns of V are formed by the eigenvectors of
, while the columns of Uareformedbytheeigenvectors
of A . Singular values in S are also square roots of
eigenvalues fromA or …singularvaluesarealways
numbers that are genuine.
It is normally acts on the host image, or thehostimageis
first segmented into numerous small blocks and then
decomposed using SVD to generate singular values, which
are used to incorporate watermark information. The SVD
coeffects’ size is constant, the singularvaluescandescribean
image’s essential algebraic properties, and singular values
are likely to vary substantially when the image is slightly
disrupted.

3.4 Encryption
Encryption is essential to Internet security today. The
encryption system transformsandencryptssensitivedata as
Code with mathematical calculations. only the correct key
reveals the original data and make sure it's safe from
everyone Except authorized parties. encryption is Hide data
by transforming it into a secret code.
Data encryption and decryption research known as
cryptography. For computers, what is plaintext? Ciphertext
refers to encrypted data, but unencrypted data.
There are three main types of encryptions for security,
these are AES, DES, and RSA. Advanced EncryptionStandard
(AES) is best Encryption standard. AES Encrypts data in
single blocks rather than as individual bits. the Block size
indicates the name of each AES type of encrypted data. Here,
same key is used for encryption and decryption. Decryption,
although the technique is implemented Separately.
This process involves encryption and decrypt, but the
only difference is the decryption Implemented in reverse
order.
=AES ( )
Where, is the previous encrypted block and is the
encryption key.
3.5Compression
Image compression is a technique for diminishing the
irrelevance and redundancy of image data so that it can be
stored or transmitted in a more Efficient manner. Image
compression is the process of Shrinking the size of an image
without sacrificing its quality.
In a file and send or communicate with others. One of
the most widely used transform techniques for image
compression of medical images using wavelets is Discrete
Wavelet Transform (DWT). This DWT is very useful for
compressing signal and shows better results for medical
gray scale images.
While using DWT the important parameters that are
taken into consideration are testing the image, wavelet
function, number of iterations and calculation complexity.
These wavelets transforms are used to process and improve
signals in fields like medical imaging where image
degradation is not tolerated.
4. PROPOSED WORK
Fig -2: Block diagram of proposed method
The datasets considered in this project are a set of retina
images and MRI of brain images.
4.1 Watermark Embedding:
Here, we design an algorithm for the application of
the watermark.watermark, encryption,andcompressionare
represented in this algorithm. In the process of watermark
embedding, two level DWT is applied on the host image and
the low frequency sub band was then obtainedandsubband
was further processed by HD and SVD. In addition, the
scaling factor is added to control the watermark embedding
strength. The new value was subdivided once again to get a
new singular value which was used toreconstructfrequency
sub band. Finally, the watermarked image was formed by
making using of new sub band after the process of inverse
discrete wavelet transform. Afterthatapplythecompression
on the watermarked image and the AES encryption is
applied to the compressed image. The decryption and the
decompression are conducted separately.
Algorithm for applying watermarking to an image:
1. Start the program
2. Look at the first image in the input (cover image).
3. Open Dataset 2 (Host image) or image 2 (input
image).
4. Resize the image 1 to 300X300.
5. Resize the image 2 to 300X300 and generate a key
value to apply wavelets on the images.
6. To partition the image for watermarking, use the
DWT approach using the haar wavelet.
7. After applying the DWT to the image, it is separated
into four sub-bands: h-LL, h-LH, h-HL, h-HH.
8. HD is performed on LL, and it is shown as PHP T =
HD(LL).
9. Extract the RGB color from input image 1 and input
image 2, apply SVD algorithm on red, green, and
blue color and SVD to H i.e., HUw HSwHVT
w=SVD(H).
10. W is applied with SVD, i.e., UwSwVT
w = SVD(W).
11. Compute an embedded singular value HS∗
w by
adding HSw and Sw with a scaling factor α by HS∗
w =
HSw + αSw

12. The watermarked sub-band H∗ is generated by
using the inverse SVD, i.e., H∗ = HUw HS∗
w HVT
w .
13. Apply the watermarking with some intensity and
get the output image with the help of IDWT.
14. Save the watermarked image with the help of
imwrite command.
15. Apply the compression on the watermarked image
and save the compressed image in a secret folder
with the help of imwrite command.
16. Apply the encryption to the compressed
watermarked image.
4.2 Watermark Extraction:
The operation of DWTwasappliedtothewatermarked
image, and the low frequency approximate coefficient of
Which was further decomposed. The Watermark was
obtained under the utilization of the original host image and
the newly formed singular value.
Algorithm for de-watermarking:
1. Start the program
2. Now Read the secret folder.
3. Save the Decrypted & decompressedimagewiththe
help of imwrite command.
4. Read the watermarked image.
5. Apply the DWT technique with Haar wavelet to
divide the image for watermarking. The image is
then divided into 4 sub-bands i.e., ωm-LL, ωm-LH,
ωm-HL, ωm-HH.
6. HD is performed on LLw by PwHwPT
w = HD(LLw)
Apply SVD to Hw, i.e., HU∗
wHSb∗
wHV∗
w
T = SVD(Hw)
and Extract the RGB color from original image and
apply SVD algorithm on red, green, and blue color.
The extracted singular value S∗
w is gained by
S∗
w = (HSb∗
w − HS∗
w)/α. The extracted watermark
W∗ is reconstructed by inverse SVD, which is
described by W∗ =Uw2S∗
wV-T
w2 .
7. Apply the watermarking with some intensity and
get output image with help of IDWT.
8. Save the extracted watermark with the help of
imwrite command.
9. Calculate the parameters like MSE,PSNR,RMSEand
CR of watermarked image.
5. EXPERIMENTAL RESULTS
In this section, the performance measures were
evaluated and proposed algorithm is evaluated using the
available dataset. which are based on the TCGA and _test
datasets is presented in this work. The TCGA dataset
contains 50 images, _test dataset contains
25 images.
Table -1: Database Description
S. No Dataset
type
Size of
dataset
No. of images
1 TCGA 14MB 50
2 Test
dataset
6MB 25
5.1 Analysis of algorithm
Many of watermarking algorithms are unable to provide
better quality and imperceptibility of the image have been
subjected to performance measures. To address this issue,
this work is to perform DWT-HD-SVD watermarking by
comparing the performance measures of the watermarked
image.
Watermarking when using Brain MRI and Retina
images
Fig -5(a): Input images for watermarking
The cover image is at the left, at the same time as the
watermark image is at the right, as shown in fig.5(a). When
we apply the watermark embedding to the MRI of brain
database all the images are hidden under cover image but
earlier than making use of the watermark the pre-processing
level is involved.
Fig -5(b): input images after preprocessing
As illustrated in fig 5(b), the input images are rescaled to a
consistent size.

Fig -5(c): application of wavelets to input images
The wavelets are applied to both the input image
and the watermarked image after the pre-processing stage,
dividing the images into four sub-bands: LL-1, LH-1, HL-1,
HH-1, LL-2, LH-2, HL-2, and HH-2. Here, the LL sub-bands
containing the most information.
Fig -5(d): watermarking and encryption
As illustrated in fig.5(d) the left image shows the
watermarked image after thewatermarkingphasetheimage
is subjected to compression and encryption. The encrypted
image will be like shown in the above rightmost figure.
Fig -5(e): Decrypted and extracted image
The decrypted image is generated by applying the
decryption to thecompressedimage,asshowninfig.5(e).The
extracted watermarked image is presented on the right. By
applying the De-watermarking, the Brain MRI image was
retrieved from the watermarked image.
Image quality measurement:
Some quality of measures can be used to determine
the distortion in the watermarked image by comparing it to
the original image. The following sections are widely used.
Mean Square Error (MSE):
It is the averaged value of the square of the pixel-by-pixel
difference betweentheoriginalimageandstego-imagewhich
gives us a measure of the error produced in that cover image
by using the data embedding process.
Where P, Q represents the height and width of the image
respectively.
• m (i, j) represents the original images pixel
value and
• n (i, j) represents the pixel values of embedded image.
Peak Signal to Noise Ratio (PSNR):
To find the watermarked image’s quality loss in
comparison to the original image. The PSNR of an image
affects its imperceptibility. PSNR is calculated by using the
formula as follows.
PSNR = 10 log10 = (L*L/MSE)
Here, L is the value of the image height. For 8-bit image,
L=255. Any image with a brightness of greater than 30 DB is
acceptable in the application of multimedia. In medical
imaging the data quality is paramount, and PSNR of around
50Db indicates that the image is of highqualityandthatthere
has been no significant degradation in the image
compared to the original.
Root Mean Square Error (PMSE):
The squared root of MSE is used to calculate RMSE. It is a
metric used to indicate howmuch apixelchangesinmeansof
processing. Because RMSE is dependent of scale, it should
only use to evaluatethe prediction errors of differentmodels
for a single variable, not between the variables.
Where, f=forecasts (unknown results) 0=observed values
(known results)
Compression Ratio (CR):
The squared root of MSE is used to calculate the Root
Mean Square Error (RMSE). The Root Mean Square Error
Image compression is the process of reducing the size of an
image file in bytes without sacrificing the images’ quality.
More images can be saved in each amount of disc or memory
space because of the smaller file size.

CR= n1/n2
Where, let n1, n2 denotethe number of bitsintheoriginal
and compressed images
Fig -5(f): Performance measures for the MRI Database
This figure shows the performance measures and graph
for the Brain MRI database and the time taken for the entire
watermarking and de-watermarking is around 5 sec. The
encrypted and compressed images are stored in a separate
folder.
Chart -1: Performance metrices for the Brain MRI Dataset
using the db1 wavelet filter
Chart -2: MSE for Brain MRI for different types of orders of
dB filters
This MSE figureshows the result fortheMSEvaluesofthe
Brain MRI dataset for different orders of the Daubechies
wavelet filters. The Mean Square Error (MSE) is used to
determine the degradation of image.
Chart -3: PSNR for brain MRI for different types of orders
of dB wavelet filters
This figure shows the results for PSNR values of the Brain
MRI dataset for different orders of the Daubechies wavelet
filters. To determine the imperceptibility, PSNR is calculated
to the extraction of watermarked images.
6. CONCLUSIONS
In this paper, the proposed image watermarking system
is based on DWT-HD-SVD transforms. The invisibility and
robustness of this method are analyzed by the numerical
simulation trails and the results shown the better quality.
In addition, the comparison with the related works is
listed and the corresponding metric values show that the
proposed method has a better performance which is used to
secure the images transferred via telemedicine as much as
possible. These images are watermarked with the patient’s

details to avoid any mistake between the patient’s
radiographs. As a result, during the extraction, the doctor
will be able to check with certainty thatthereports belongto
the treated patient.
REFERENCES
[1] D. Bouslimi, G. Coatrieux, M. Cozic, and C. Roux, “A
joint encryption/watermarkingsystemforverifying
the reliability of medical images,” IEEE Trans. Inf.
Technol. Biomed., vol. 16, no. 5, pp. 891–899, Sep.
2012.
[2] M. N. A. Wahid, A. Ali, B. Esparham, and M. Marwan,
"A comparison of cryptographic algorithms: Des,
3DES, AES, RSA, and blowfish for guessing attacks
prevention," J. Comput. Sci.Appl.Inf.Technol.,vol.3,
pp. 1–7, Aug. 2018.
[3] S. Haddad, G. Coatrieux, A. Moreau-Gaudry, and M.
Cozic, "Joint Watermarking-Encryption-JPEG-LSfor
Medical Image Reliability Control in Encrypted and
Compressed Domains," in IEEE Transactions on
Information Forensics and Security, vol. 15, pp.
2556-2569, 2020, DOI:
10.1109/TIFS.2020.2972159.
[4] Sumedh P. Ingale1, Prof. C. A. Dhote,” Digital
Watermarking Algorithm using DWT Technique,”
International Journal of Computer Science and
Mobile Computing, Vol.5 Issue.5, May- 2016,pg.01-
09
[5] Sunesh, Vinita Malik, Neeti Sangwan, Sukhdip
Sangwan.” Digital Watermarking using DWT-SVD
Algorithm,” Advances in Computational Sciences
and Technology ISSN 0973-6107 Volume 10,
Number 7 (2017) pp. 2161-2171 © Research India
Publications
[6] Thakkar, F.N. and Srivastava, V.K., 2017. A blind
medical image watermarking: DWT-SVD based
robust and secure approach for telemedicine
applications. Multimedia Tools and Applications,
76(3), pp.3669-3697.
[7] W. Puech and G. Coatrieux. Chapter 10: Coding:
Encryption Watermarking-CompressionforMedical
Information Security. Compression of Biomedical
Images and Signals, A. Na¨ıt-Ali and Christine
CavaroMenard, Digital Signal Processing, ISTE-
Wiley, May 2008.
[8] Puech, W. (2008). [IEEE 2008 First Workshops on
Image Processing Theory, Tools and Applications
(IPTA) - Sousse, Tunisia (2008.11.23- 2008.11.26)]
2008 First Workshops on Image ProcessingTheory,
Tools and Applications - Image Encryption and
Compression for Medical Image Security. , (), 1–2.
doi:10.1109/ipta.2008.4743800 R. Nicole, “Title of
paper with the only first word capitalized,” J. Name
Stand. Abbrev., in press.
[9] S. Haddad, G. Coatrieux, A. Moreau-Gaudry, and M.
Cozic, "Joint Watermarking-Encryption-JPEG-LSfor
Medical Image Reliability Control in Encrypted and
Compressed Domains," in IEEE Transactions on
Information Forensics and Security, vol. 15, pp.
2556-2569, 2020, DOI:
10.1109/TIFS.2020.2972159.
[10] S. P. Metkar and M. V. Lichade, “Digital image
security improvement byintegratingwatermarking
and encryption technique,” in Proc. IEEE Int. Conf.
Signal Process., Computer. Control (ISPCC), Sep.
2013, pp. 1– 6.
[11] Sumedh P. Ingale1, Prof. C. A. Dhote,” Digital
Watermarking Algorithm using DWT Technique,”
International Journal of Computer Science and
Mobile Computing, Vol.5 Issue.5, May- 2016,pg.01-
09
[12] Sunesh, Vinita Malik, Neeti Sangwan, Sukhdip
Sangwan.” Digital Watermarking using DWT-SVD
Algorithm,” Advances in Computational Sciences
and Technology ISSN 0973-6107 Volume 10,
Number 7 (2017) pp. 2161-2171 © Research India
Publications
[13] Thakkar, F.N. and Srivastava, V.K., 2017. A blind
medical image watermarking: DWT-SVD based
robust and secure approach for telemedicine
applications. Multimedia Tools and Applications,
76(3), pp.3669-3697.
[14] W. Puech and G. Coatrieux. Chapter 10: Coding:
Encryption Watermarking-CompressionforMedical
Information Security. Compression of Biomedical
Images and Signals, A. Na¨ıt-Ali and Christine
CavaroMenard, Digital Signal Processing, ISTE-
Wiley, May 2008.
[15] W. Puech and J.M. Rodrigues. A New Crypto-
Watermarking Method for Medical Images Safe
Transfer. In Proc., 12th European Signal Processing
Conference (EUSIPCO'04), pages 1481–1484,
Vienna, Austria, 2004.

Hybrid Transforms Based Watermarking, Encryption and Compression of Bio-Medical Images

More Related Content

Similar to Hybrid Transforms Based Watermarking, Encryption and Compression of Bio-Medical Images (20)

More from IRJET Journal (20)

Recently uploaded (20)

Hybrid Transforms Based Watermarking, Encryption and Compression of Bio-Medical Images