3rd IEEE International Conference on Electronic Engineering, Menoufia University, Egypt. ICEEM2023, 7-8 Oct.

2023

A Software Design and FPGA Implementation of

Telemedicine Security System
2023 3rd International Conference on Electronic Engineering (ICEEM) | 979-8-3503-2351-1/23/$31.00 ©2023 IEEE | DOI: 10.1109/ICEEM58740.2023.10319582

Hany Y. Ahmed Taha Elsayed Taha Adel S. El-Fishawy

Electronics and Electrical Electronics and Electrical Electronics and Electrical
Communications Engineering Communications Engineering Communications Engineering
Department Department Department
Faculty of Electronic Engineering, Faculty of Electronic Engineering, Faculty of Electronic Engineering,
Menoufia University Menoufia University Menoufia University
Menouf, Egypt 32952,Menouf, Egypt Menouf, Egypt
[email protected] [email protected] [email protected]

Emad A. Elshazly
Engineering Department
Nuclear Research Center, Egyptian
Atomic Energy Authority
Cairo, Egypt.
[email protected]

Abstract—Secure data transfers in communication data. One of the most commonly accepted data hiding
applications for telemedicine play a significant role in approaches is image data hiding [2, 3].
security. In this paper, a secure telemedicine corneal data
concealment algorithm based on the pixel segmentation The two categories of data hiding strategies are transform
strategy (PSS) and generalized exploiting modification domain and spatial domain. One spatial domain strategy for
direction (GEMD) is presented. The suggested algorithm is concealing sensitive information in digital photographs is the
implemented using the Xilinx System Generator (XSG) and exploiting modification direction (EMD) algorithm. To
hardware/software co-simulation is carried out using the incorporate a significant payload and produce a satisfactory
Spartan 3E FPGA board, which offers the least embedding stego-image quality, various EMD algorithms are applied. In this
time and offers a quicker, programmable, and commercial paper, a method was put out to address the shortcomings of
hardware solution for secure communication. The proposed earlier data masking techniques. An embedding phase and an
approach can be applied in low-cost embedded system extraction phase make up the suggested approach. In each of
applications since it has high embedding speed and low them a block of pixels in the cover image is divided into two
computational complexity. The outcomes of the practical components during the hiding phase and extracting phase of the
tests of the suggested algorithm are the same as the proposed algorithm: the first component is the vector of
simulation results of the same algorithm. The suggested coordinate area (VCA) while the second component is the vector
algorithm has a high payload capacity while maintaining modification area (VMA). Each block's VCA vector is left
good stego-image quality in comparison to the earlier data
unaltered, while the VMA vector is adjusted.
concealment algorithms. The proposed algorithm presents a
practical results as the same as that of simulated results. The main contribution of this paper is to propose a more
secure data hiding and extracting algorithms based on PSS and
Keywords—Telemedicine Data Hiding, PSS, GEMD, GEMD scheme in telemedicine applications and also presented
FPGA and XSG. FPGA implementation based on XSG of the proposed
I. INTRODUCTION algorithms and compare the practical results with the simulation
results in previous work. The suggested algorithm distributes
Due to the significant advancements and quick growth in the the consequent VCA vector (which contains the VCA of each
realm of technology, the internet has emerged as the primary block pixel) uniformly according to "n" and indicates the virtual
medium for user-to-user data transmission, sharing, and adjustment on pseudorandom vectors in n-dimensions for
exchange. For private data transfer, secret data archiving, data carrying the secret data, where n = LVMA-1 and LVMA is the
alteration prevention, and access control systems for digital length of the consequent VMA vector [4–7]. The hidden
content distribution, telemedicine security is crucial. One of the information is taken out of the stego-image's pixels during the
algorithms used to secure data is data hiding. Digital extraction method. The embedding and extracting methods are
watermarking and steganography are its two basic methods [1]. created and implemented utilizing FPGA via XSG in this paper's
Digital watermarking is mostly used to safeguard electronic suggested telemedicine security system [8, 9]. The Spartan 3E
products' copyrights, but steganography is a technique for hidden FPGA board is used for the hardware/software co-simulation [3,
communication. Steganography's main objective is to covertly 10–12].
transmit data by hiding the information's existence within
another kind of media. Cover objects are the materials used to The rest of this paper is structured as follows: The proposed
conceal data. Stego-object refers to the cover and the concealed telemedicine corneal data hiding algorithm, based on PSS and

979-8-3503-2351-1/23/$31.00 ©2023 IEEE 1 https://iceem2023.conferences.ekb.eg

Authorized licensed use limited to: SRM University Amaravathi. Downloaded on November 04,2024 at 11:53:01 UTC from IEEE Xplore. Restrictions apply.
 3rd IEEE International Conference on Electronic Engineering, Menoufia University, Egypt. ICEEM2023, 7-8 Oct. 2023

GEMD, is introduced in Section III. Section IV displays the VMA2, the VCA vector is made up of VCA1 and VCA2
FPGA implementation of the proposed hiding and extracting [7].
algorithms using XSG. Section V presents the FPGA H/W co-
simulation. Section VI presents the findings of actual
experiments and discussions. The conclusion is presented in
section VII.
II. LITERATURE SURVEY
On the basis of EMD algorithms, several image information
concealing algorithms have been created. Based on the original
EMD technique, Zhang et al. presented an information
concealing approach [4]. When utilizing this method to hide
information in digital images, where, each block of "n" pixels in
the original image carries a secret number in the (2n+1)-ary
notational system. By utilizing the relationship between "n"
neighboring pixels of the cover image, the EMD method
embeds hidden data. Each block affects just one of the two Fig. 1. The proposed telemedicine data hiding and extracting algorithms
adjacent pixels it contains when there are two adjacent pixels in (a) Hiding Phase, and (b) Extracting Phase.
a block, either by deleting 1, adding 1, or keeping the pixel
unaltered. From a geographical perspective, there are five
possible outcomes in this scenario: travelling downward,
upward, left, right, or not moving at all. The EMD approach
maintains acceptable stego-image quality while increasing the
capacity of secret data [5]. An improved exploiting modification
direction (IEMD) technique was put out by Lee et al. [6]. The
embedding capacity of this approach is 1.5 times higher than that
of the EMD algorithm. More details about the EMD working
principles can be found in [6]. Fig. 2. The PSS working principles.

By suggesting an information hiding method based on a B. The GEMD Telemedicine Data Hiding Algorithm
vector of coordinates and a strategy of pixel segmentation, they
More details about the GEMD working principles can be
also improve the EMD's ability to store secret data. The Vector
found in [6]. If the MSBs rather than the LSBs are changed, the
Modification Area (VMA) space was discovered to be
distortion of the cover image increases significantly.
insufficient, and the transmitter and receiver had to exchange
Additionally, it is much simpler to identify the MSBs that have
variable codes before they could continue to interact. The
been modified. As a result, only the LVMA LSBs dispersed over
GEMD approach was published by Kuo et al. where each (n+1)-
the VMA of the specified pair of original pixels are potentially
bit block of the binary secret data stream is hidden in "n"
changed in the suggested method. The function of extraction
consecutive pixels block of the cover image according to the
f(g1, g 2, …, g n) is [6]:
basic GEMD technique; each pixel may have 1 added to it,
deleted from it, or kept unchanged [6].
 n 
III. THE PROPOSED TELEMEDICINE DATA HIDING AND f GEMD ( g1 , g 2 ,...., g n ) =   g i . (2 i − 1)  mod( 2 n +1 ) (1)
 i =1 
EXTRACTING ALGORITHM
Lee et al.’s information hiding algorithm improved the The suggested embedding algorithm's steps are outlined as
capacity of secret information to be hidden using original EMD follows:
and PSS. Since both the transmitter and the receiver must Inputs Ic, the telemedicine original image and S, the
initially have the same code table, this technique does not have
secret information stream
enough room for utilization. Telemedicine information hiding
Outputs Is, the stego-image
algorithm based on PSS and GEMD data hiding method is
proposed in order to remove these drawbacks. Figure (1) shows Step 1: Determine the values of VMA and VCA for each
the proposed telemedicine data hiding and extracting pair of pixels (p1, p2) in the telemedicine original
algorithms. The proposed algorithm is performed practically image, and determine the length of VMA vector
using Xilinx system generator. LVMA.
Step 2: Calculate "n", where n = LVMA - 1.
A. Pixel Segmentation Strategy
Step 3: Create the (g1, g2,..., gn) generating vectors by
In PSS, a pair of telemedicine cover image pixels are allocating the VCA vector according to "n".
used to incorporate hidden data. Figure (2) illustrates the Step 4: Use Equation (1) to calculate the embedding
PSS working principles, where each pair of grayscale function t = f(g1, g2,..., gn).
pixels can be divided into four segments. These segments Step 5: Estimate the value of d = s – t.
are labeled VCA1, VMA1, VCA2, and VMA2, respectively.
The MSBs of the first and second pixels, respectively, of Step 6: Covert "d" into binary and before filling out the
the telemedicine original image are represented by VCA1 VMA.
and VCA2. The LSBs of the first pixel and the second pixel Step 7: Put the blocks back in their proper positions to
of the cover image respectively, are represented by VMA1 reassemble the telemedicine stego 2-pixels (p′1,
and VMA2. The VMA vector is made up of VMA1 and p′2).

979-8-3503-2351-1/23/$31.00 ©2023 IEEE 2 https://iceem2023.conferences.ekb.eg

Authorized licensed use limited to: SRM University Amaravathi. Downloaded on November 04,2024 at 11:53:01 UTC from IEEE Xplore. Restrictions apply.
 3rd IEEE International Conference on Electronic Engineering, Menoufia University, Egypt. ICEEM2023, 7-8 Oct. 2023

C. The GEMD Telemedicine Data Extracting Algorithm strategy and GEMD embedding and extracting methods is
By following the directions in the extracting method, implemented using XSG blocksets [10].
the secret data concealed in the cover image can be
retrieved from the telemedicine stego-image. As with the
embedding technique, the extraction procedure must divide
a given pair of stego-pixels into the VMA and VCA [6].
Fig. 3. Design flow of hardware implementation of the suggested
 n  technique.
f GEMD ( g , g 2 ,...., g n ) =   g 'i . (2 i − 1)  mod( 2 n +1 )
'
1
' '
(2)
 i=1  A. The pre-processing stage
The suggested extracting algorithm's steps are outlined as In the instance of the suggested telemedicine data
follows: hiding approach, the pre-processing phase is carried out
using MATLAB, which supplies the inputs to the Xilinx
Inputs Is, telemedicine stego-image blocks as a particular vector array suited for FPGA
Outputs S, stream of secret data bitstream compilation as follow:
Step 1: The telemedicine stego-image's block (p'1, p'2) • Convert a 2-D image to a 1-D one: This is done by
should be split into VMA and VCA sections for breaking a 2-D image down into a single vector of
each block. pixels. Two vectors are obtained; one with even
Step 2: Determine the length of VMA vector LVMA. indices and the other with odd indices are
Step 3: Calculate "n", where n = LVMA - 1. segregated from this single vector.
Step 4: Create "n" generating vectors, designated as (g1,
• Text: In this step, the text also is prepared in format
g2, … , gn), by allocating the VCA vector
where the Xilinx blocks can read it.
according to n.
Step 5: Estimate (g′1, g′2, … , g′n) using VMA, from (g1, While the pre-processing phase for the proposed
g2, … , gn) by follow the GEMD method and telemedicine extracting algorithm is also done using
Table (1). MATLAB, the same as in image pre-processing phase of
Step 6: Estimate and convert the secret information s = data hiding algorithm.
f(g′1, g′2, … , g′n) obtained from Eq. (2) into • Convert a 2-D image to a 1-D one: This is done by
binary. breaking a 2-D image down into a single array of
Step 7: To get the hidden secret data stream S, pixels. Two vectors are also obtained; one with
concatenate the secret data "s" that is hidden in even indices and the other with odd indices are
each block. segregated from this array.
TABLE I The Pixel adjusted quality according to "d" [8]. B. The proposed telemedicine data hiding and extracting
algorithms
The content in the VMA vector and VCA vector for
each pair of pixels of the original image determines how
the suggested telemedicine data concealment strategy is
implemented. The VMA vector is just 3 bits, while the
VCA vector of each pair of pixels in the suggested
algorithm is 13 bits. Figure (4) depicts the model-based
design for the suggested data hiding approach for the
aforementioned instance utilizing Xilinx blocksets.
The proposed telemedicine data hiding approach is
implemented in accordance with the length of bits in both
VMA and VCA coordinate vectors for each pair of pixels
IV. FPGA IMPLEMENTATION OF THE SUGGESTED DATA in cover image. In this instance, (VCA1, VMA1) = (7, 1)
HIDING AND EXTRACTING ALGORITHMS and (VCA2, VMA2) = (6, 2). Figure (4a) depicts the XSG
model-based design utilizing Xilinx blocksets to simulates
The implementation content of the suggested
the suggested data hiding steps in the proposed approach
telemedicine information hiding and extracting algorithms
for the aforementioned instance, while Figure (4b) presents
based on pixel segmentation strategy and GEMD
the construction of the subsystem in Figure (4a). Each step
embedding and extracting methods using Xilinx blocks
in the proposed hiding approach is implemented using
and Simulink blocks is investigated through three stages;
Xilinx blocksets.
the pre-processing stage, the proposed telemedicine
information hiding or extracting algorithm and the post- The model-based design of the suggested telemedicine
processing stage [10]. Figure (3) shows the design flow for data extracting algorithm is shown in Figure (5). Figure
the hardware implementation of the suggested (5a) presents the XSG model-based design utilizing Xilinx
telemedicine data hiding and extracting methods. blocksets to simulates the suggested data extracting steps
MATLAB software is used in the pre- and post-processing in the proposed extracting approach, while Figure (5b)
phase design, while the proposed telemedicine data hiding presents the construction of the subsystem in Figure (5a).
or extracting algorithms based on pixel segmentation Each step in the proposed approach is implemented using
Xilinx blocksets.

979-8-3503-2351-1/23/$31.00 ©2023 IEEE 3 https://iceem2023.conferences.ekb.eg

Authorized licensed use limited to: SRM University Amaravathi. Downloaded on November 04,2024 at 11:53:01 UTC from IEEE Xplore. Restrictions apply.
 3rd IEEE International Conference on Electronic Engineering, Menoufia University, Egypt. ICEEM2023, 7-8 Oct. 2023

C. The post-processing stage

This stage is used to combine again both stego-signals
obtained from the Xilinx models and turn the resulting
vector back into a 2-D image in the event of a data
concealing technique.
• Convert a 1-D signal to a 2-D image: This is done
by combining the resulting vectors into a single
one, then transform this vector to a 2-D image,
from which the stego-image is created.
(4a)
While in case of data extracting algorithm, the post-
processing phase is done using MATLAB to construct the
extracted image and the original text.
V. THE FPGA H/W CO-SIMULATION
The system generator, the Spartan 3E kit, and JTAG
cable are used to generate a new hardware Xilinx block
containing the above XSG based model for embedding
algorithm shown in Figure 4 and extracting algorithm
shown in Figure 5. To conform to the needs of the
underlying hardware board, a model can be co-simulated.
(4b) It is now possible to confirm that the obtained results using
Fig. 4. The FPGA implementation of the proposed GEMD data hiding hardware and the obtained results using software are
algorithm. identical. Figure (6) displays the hardware co-simulation
for the suggested telemedicine data hiding (6a) and
extraction (6b) techniques.

(5a)

(6a)

(6b)
Fig. 6. The hardware co-simulation for the proposed telemedicine data
hiding and extracting algorithms.

VI. RESULTS AND DISCUSSION OF PRACTICAL EXPERIMENTS

A. Database
On various telemedicine corneal images, the proposed
data hiding and extracting algorithms have been used, and
a comparison study of the algorithms is conducted. Images
of the corneal cover are displayed in Figure (7).
B. Performance Measurement
The MSE and the PSNR in (dB) are the performance
evaluation metrics used in all tests. To evaluate the
embedding payload capacity as well as visual telemedicine
(5b) stego-image quality of the suggested technique, the
technique is carried out using MATLAB program on the
Fig. 5. The FPGA implementation of the proposed telemedicine data
extracting algorithm.
Intel(R) Core (TM) i7-10870H CPU @ 2.20GHz CPU

979-8-3503-2351-1/23/$31.00 ©2023 IEEE 4 https://iceem2023.conferences.ekb.eg

Authorized licensed use limited to: SRM University Amaravathi. Downloaded on November 04,2024 at 11:53:01 UTC from IEEE Xplore. Restrictions apply.
 3rd IEEE International Conference on Electronic Engineering, Menoufia University, Egypt. ICEEM2023, 7-8 Oct. 2023

computer with 16 GB RAM, also FPGA implemented 1 N −1 M −1

using XSG. In comparing the performance of the MSE =
M ×N
  [I S (i, j ) − I C (i, j )]2 (3)
suggested data hiding technique, the hiding rate is R = i =0 j =0

S/(M×N), to compute the length of bits concealed by one

PSNR = 10 log10 ( f max
2
/ MSE ) (4)
pixel of the original image, i.e., the hiding rate (bpp),
where S is the length of bits of the secret data to be Where M and N are the telemedicine original images'
embedded, and (M×N) is the total length of pixels of the respective row and column numbers, and fmax is the highest
original image. Table (2) compares the hiding rate (bpp) value from the original image.
and the payload capacity of the existing EMD techniques
and the suggested one. A measure that can be used to gauge how similar (or
dissimilar) the stego-image and the telemedicine original
image are is the Normalised Cross-Correlation (NCC). The
NCC is determined as shown below [3]:
N −1 M −1
  I C (i , j ) × I S (i , j )
i =0 j =0
NCC = N −1 M −1 (5)
  I C (i , j )
2

i =0 j =0

The hidden rate and PSNR for the suggested technique

(simulation and practical) are listed in Table (3), where the
hidden rate means how many bits are hidden in one pixel.
For (LVMA1, LVMA2) = (1, 2), this means that three bits can
Fig. 7. The telemedicine corneal cover images. be hidden in two pixels; i.e 1.5 bits in each pixel. If the
hidden rate is increased more than 1.5 bpp, the PSNR
The payload capacity is 1.16×M×N bits and the hiding decreased and vice versa. The MSE, PSNR, and NCC of
rate is given as log2(5)/2 = 1.16 bpp, when n = 2 in the the suggested embedding techniques are contrasted in
original EMD hiding process. For the same pair of pixels, Table (4). As can be seen, the PSNR = 50.288 dB, when
the IEMD hiding technique has a 1.5 bpp hiding rate and a the hidden rate is 1.5. Figure (8) displays the associated
1.5×M×N bits payload capacity. When a block of pixel telemedicine stego-images obtained by following the steps
pairs is utilized in the pixel segmentation strategy and the of the proposed data hiding algorithm which is
EMD embedding process and LVMA = 3, the hidden rate is implemented using Xilinx blocksets. Figure (8) presented
half of log2(2(2 LVMA-1 – 1) + 1) = 1.4 bpp, and the also the PSNR values of the telemedicine stego-images.
corresponding payload capacity is 1.4×M×N bits. When n
= 2, the hidden rate in the GEMD embedding algorithm is TABLE III The hiding rate and PSNR suggested algorithms (simulation
3/2, or 1.5 bpp, and the corresponding payload capacity is and practical) [13].
1.5×M×N bits.

TABLE II Comparison between existing EMD techniques and the

suggested one [7, 8].

TABLE IV The hiding rate, MSE, PSNR and NCC of the proposed
algorithms.

When each pixel-pair is utilized in the suggested

embedding algorithm and LVMA = 3, the hidden rate is 1.5
bpp and the corresponding payload capacity is 1.5 times
M×N bits; however, as the hidden rate rises, the payload
capacity also rises.
The two error measures utilized to assess the
effectiveness of the aforementioned techniques are the
MSE and the PSNR, in dB. The MSE measures the peak
error, whereas the PSNR measures the total squared error
between the telemedicine corneal original image and the
telemedicine corneal stego-image [3].
The following equations are used to determine the
MSE and PSNR, where IC and IS are the telemedicine
original and stego images, respectively [3, 5, 6]:

Fig. 8. The corresponding telemedicine stego-images and their PSNR

values.

979-8-3503-2351-1/23/$31.00 ©2023 IEEE 5 https://iceem2023.conferences.ekb.eg

Authorized licensed use limited to: SRM University Amaravathi. Downloaded on November 04,2024 at 11:53:01 UTC from IEEE Xplore. Restrictions apply.
 3rd IEEE International Conference on Electronic Engineering, Menoufia University, Egypt. ICEEM2023, 7-8 Oct. 2023

VII. CONCLUSIONS [5] Xinpeng Zhang, Shuozhong Wang, Efficient

Steganographic Embedding by Exploiting Modification
The algorithms for hiding and extracting corneal data Direction, IEEE Communications Letters 10(11):781 – 783,
from telemedicine have been suggested in this paper and November 2006.
were due to the PSS and GEMD embedding method. The [6] W. C. Kuo, and C. C. Wang. Data hiding based on
embedding rate of the suggested approach is higher than generalized exploiting modification direction method.
the GEMD embedding method. XSG is used to implement Imaging Science journal, Vol. 61, No. 6, pp. 484 – 490,
2013.
the embedding and extracting phases, while Spartan 3E
[7] YANXIAO LIU, CHINGNUNG YANG, AND QINDONG
FPGA kit is used for the hardware software co-simulation. SUN, Enhance Embedding Capacity of Generalized
The outcomes of actual tests for the suggested technique Exploiting Modification Directions in Data Hiding, IEEE
can conceal a larger payload than the earlier EMD ACCESS, Volume 6, 2018.
embedding algorithms while maintaining a reasonable [8] W. C. Kuo and S. Y. Chang, Hybrid GEMD Data Hiding,
degree of stego-image quality (50.288 dB). The outcomes Journal of Information Hiding and Multimedia Signal
demonstrate that the suggested technique is one of the best Processing, Vol. 5, No. 3, pp. 420-430, July 2014.
data concealing and extracting algorithms since it produces [9] W. Luo, F. Huang and J. Huang, A More Secure
Steganography Based on Adaptive Pixel-value Differencing
high embedding payload capacities and great stego-image Scheme, Multimedia Tools and Applications, Springer, Vol.
quality. It can be concluded that the suggested approach is 52, No. 2-3, pp. 407-430, April 2011.
superior to existing methods for hiding and extracting [10] N. P. Rautand and A. V. Gokhale, FPGA Implementation
image data. The proposed algorithm presents a practical for Image Processing Algorithms Using Xilinx System
results typically as that of simulated results. Generator, IOSR Journal of VLSI and Signal Processing,
Vol. 2, No. 4, pp. 26-36, 2013.
REFERENCES [11] T. Janani, M. Brindha, A secure medical image
transmission scheme aided by quantum representation,
[1] Amal Hafsa, Mohamed Gafsi, Jihène El Malek and Mohsen Journal of Information Security and Applications, Volume
Machhout, FPGA Implementation of Improved Security 59, June 2021.
Approach for Medical Image Encryption and Decryption,
Scientific Programming, February 2021. [12] Mu’Ath Al-Shaikh, Medical image content protection by
secret information hiding to support telemedicine,
[2] Gandharba Swain, Digital image steganography using
Cryptography and Security [cs.CR]. Université de Bretagne
variable length group of bits substitution, Procedia occidentale - Brest, 2016.
Computer Science 85, ELSEVIER, pp. 31 – 38, 2016.
[13] Wen-Chung Kuo and Sheng-Yi Chang, A GEMD Data
[3] E. A. Elshazly, Safey A. S. Abdelwahab, R. M. Fikry, O.
Hiding by Pixel Segmentation Strategy, Seventh
Zahran, S. M. Elaraby, M. El-Kordy. FPGA implementation International Conference on Innovative Mobile and Internet
of image steganography algorithms using generalized Services in Ubiquitous Computing, IEEE, 2013.
exploiting modification direction and pixel segmentation
strategy, 2018 35th National Radio Science Conference
(NRSC), 2018.
[4] Zeyad Safaa Younus, Mohammed Khaire Hussain, Image
steganography using exploiting modification direction for
compressed encrypted data, Journal of King Saud
University – Computer and Information Sciences 34 (2022)
2951–2963.

979-8-3503-2351-1/23/$31.00 ©2023 IEEE 6 https://iceem2023.conferences.ekb.eg

Authorized licensed use limited to: SRM University Amaravathi. Downloaded on November 04,2024 at 11:53:01 UTC from IEEE Xplore. Restrictions apply.