Tuijin Jishu/Journal of Propulsion Technology

ISSN: 1001-4055
Vol. 45 No. 3 (2024)
__________________________________________________________________________________

Facial Image Encryption Using Enhanced Xor

Operation for Secured Transimssion on
Cloud Storage
Pranali Dahiwal1, Anagha Kulkarni2
1
Research Scholar, Vishwakarma Institute of Information Technology, S.P.Pune University
2
Cummins College of Engineering for Women, Pune, India

Abstract:- The study highlights the significance of security and privacy in image transmission over cloud
platforms, especially considering the sensitive nature of personal information involved in facial image recognition.
To address this concern, the study proposes the use of image encryption, with a focus on optimising transmission
time, which is crucial for real-time facial image identification and authentication. Initially, the author segmented
the images into smaller parts, with segment sizes of 3x1, 3x2, and 3x3. This segmentation facilitates efficient
processing and transmission and post that the proposed study uses the XOR encryption technique for encrypting
images. XOR encryption is chosen for its simplicity and computational efficiency. An experiment was conducted
to evaluate the proposed method. Overall, the study contributes to the field of secure image transmission over
cloud platforms, particularly in the context of facial image recognition. By leveraging XOR encryption and
optimizing segmentation techniques, the proposed method offers a promising solution for ensuring both security
and efficiency in image transmission processes.

Keywords: AWS, Cloud, Image Encryption, XOR, Segmentation, Time Complexity.

1. Introduction
Biometric authentication systems, including face recognition, offer several advantages over conventional
password-based systems. Firstly, biometric traits are unique to each individual, making them difficult to forge or
replicate. This uniqueness enhances security as it mitigates the risk of unauthorized access due to stolen or
guessed passwords [1][2]. Additionally, since biometric traits are inherent to individuals, there is no need to
remember or manage passwords, which can be forgotten, shared, or stolen.
Furthermore, the registration process in biometric systems involves capturing and storing biometric data, which
acts as a barrier against false authentication attempts[3]. This registration step helps ensure that only authorized
individuals can access the system, enhancing security.
However, biometric authentication systems also have limitations, with one of the most significant being their
susceptibility to spoofing attacks. Spoofing attacks involve presenting false biometric information to the system
to gain unauthorized access[4]. Synthesis, where attackers fabricate biometric data, is one form of spoofing
attack[5]. Additionally, replay attacks, where captured biometric data is replayed to the system, can also be
considered spoofing attacks[6][7].
In a public cloud-based facial authentication system, the vulnerability to spoofing attacks is heightened due to the
remote nature of the system and the potential lack of physical security measures. Attackers may attempt to
intercept or manipulate biometric data as it is transmitted over the internet, increasing the risk of successful
spoofing attacks.
To mitigate the vulnerability to spoofing attacks in public cloud-based facial authentication systems, it is crucial
to implement robust security measures such as encryption of biometric data during transmission, multi-factor

462
Tuijin Jishu/Journal of Propulsion Technology
ISSN: 1001-4055
Vol. 45 No. 3 (2024)
__________________________________________________________________________________
authentication, continuous monitoring for suspicious activity, and regular updates to security protocols.
Additionally, incorporating liveness detection techniques, which verify that
the biometric sample is from a live person and not a static or synthetic representation, can help enhance the
resilience of the system against spoofing attacks.
The paper proposes a novel method for enhancing the security and efficiency of face recognition systems deployed
on public cloud platforms like Amazon Web Services (AWS). The proposed methodology integrates image
encryption techniques with the recognition process to protect against spoofing attacks on facial images used for
authentication.
Here's a breakdown of the proposed methodology:
Image Encryption: Pre-processed facial images are encrypted using various encryption algorithms, with XOR
encryption being highlighted. XOR encryption is applied to the segmented grid of the image. This encryption
process aims to protect the facial features extracted from the images, making it difficult for attackers to manipulate
or spoof the images.
Segmentation Approach: The facial images are divided into segmented grids (e.g., 3x1, 3x2, 3x3) before
encryption. This segmentation approach helps in enhancing the security and integrity of the encrypted images. It
ensures that even if a portion of the image is compromised, the entire image remains protected, thereby reducing
the risk of successful spoofing attacks.
Integration with Recognition Process: The encrypted facial images are then used in the recognition process.
During authentication, the classifier needs to correctly identify and decrypt the submitted input using the
appropriate encryption grid and techniques. This integration ensures that only legitimate users with access to the
correct encryption keys can gain authentication.
Time Complexity Analysis: The paper evaluates the time complexity of the encryption and decryption processes.
It demonstrates that the proposed method incurs minimal overhead in terms of processing time, ensuring that the
authentication process remains efficient even with the added encryption layers.
The main contribution of the paper lies in its innovative approach to image authentication, which leverages
segmentation and advanced encryption techniques specifically tailored for cloud platforms. By integrating
encryption directly into the recognition process, the proposed methodology aims to enhance the security of facial
authentication systems against spoofing attacks while maintaining low-time complexity for encryption and
decryption operations.
Overall, this paper presents a promising avenue for improving the security of face recognition systems deployed
on public cloud platforms, addressing the critical need for robust authentication mechanisms in cloud-based
environments.
The outline of our paper is arranged as follows: Section 2 contains the literature review. Section 3 focuses on the
methodologies used in the proposed work. We detail the results of our proposed system in Section 4. Finally,
Section 5 refers to the conclusion of the paper.
2. Related Work
In the security system of digital information, especially on cloud platforms, encryption technology is a very
common technique and method. Encryption technology can be used to encrypt the original data. If the security
and reliability of the encryption method are high enough, then the security of digital information can be
protected[8][9][10][11]. Therefore, research on digital image encryption technology and methods is an important
direction for digital image security protection.
According to M. T. Elkandoz et.al., an image encryption scheme was proposed. The proposed scheme is based on
chaotic maps. The chaotic maps used in this scheme are the Arnold map, the 2D logistic sine map, and the
congruential generator. The algorithm starts by shuffling the image pixels using the Arnold map. Then, the 2D
logistic sine map and congruential generator were used to generate a 256-bit key. The shuffled image data is then

463
Tuijin Jishu/Journal of Propulsion Technology
ISSN: 1001-4055
Vol. 45 No. 3 (2024)
__________________________________________________________________________________
XORed with the 256-bit key generated from the sine map and congruential generator to produce the final
encrypted image. The proposed algorithm's performance is comparable to the counterpart schemes described in
the literature[12].
I. Younas and M. Khan proposed a scheme that is based on simplicity. They followed Shannon’s rules of confusion
and diffusion in the simplest and easiest possible way. The simplicity of the algorithm comes from the use of two
simple boolean operations. However, its strength comes from using large keys during encryption. These two
boolean operations are the XOR operation and the rotation operation. They perform these two operations on the
pixels of an image, starting with a sequential XORing operation on all of the bit pixels of the image, followed by
a circular clock-wise rotation of the bits. These two operations are repeated iteratively until the final encrypted
image is achieved. The security of the proposed method has been evaluated using three measures: key space
analysis, key sensitivity analysis, and statistical analysis. The results show that the proposed method is an effective
encryption method that can be applied in different fields of image encryption[13].
S. Kanwal et.al in the their studies proposed a new technique for the development of an s-box. The modified s-
box is based on a multiplicative group of nonzero elements of the Galois field of order 256. The proposed
technique uses exponential and Tinkerbell chaotic maps to transform a plain image into a ciphered one. Multiple
tests were performed on the proposed method, such as MSE and PSNR. The MSE value of the proposed technique
is 8053, and the PSNR value is 9.3214. The results of this paper show that the proposed s-box construction
mechanism is efficient and provides secure, real-time communications[14].
A novel approach to encrypting digital images is proposed by F. Khan et.al. The proposed scheme is based on
matrix reordering (MR), which is a scanning operation, and XOR operations. Firstly, the MR is used to perform
permutations on the bit pixels of the image. Then, a scan pattern is obtained from the MR, which describes how
the bit pixels will be rearranged. Secondly, the linear congruential generator is used to generate the pseudorandom
bit stream that will be used in the XOR operation. Finally, the rearranged bit pixels of an image are XORed with
the bit stream generated from the linear congruential generator to generate the final encrypted image. The proposed
scheme was evaluated using a histogram, correlation, cut test, dispersion test, visual testing, and speed test. The
results show that the correlation value is close to zero, which means that the relationship between the original
image and the encrypted image is very low. The cut test, dispersion test, and speed test show that the proposed
scheme can be used in real-time applications. To conclude, the proposed scheme has proven to be resistant to
statistical and differential attacks and can be used in real-time applications[15].
A new technique of image encryption is proposed by the A. Sinha and K. Singh. This technique was composed of
two main parts, which were encoding and digital signature. This proposed scheme starts by encoding an image,
the encoding is using Bose-Chaudhuri Hochquenghem (BCH), and then the digital signature of the original image
was added to the encoded version of the image. The added digital signature can be used to verify the authenticity
of the image. This proposed scheme provides three layers of security. Firstly, an error control code was used in
encoding the original image, which was very difficult to break. Secondly, the digital signature was added to the
encoded image, which verifies the authenticity of the image. Finally, there was no need to transmit the keys
separately. A digital correlation technique can be used to verify the authenticity of the decrypted image.
Transmission of digital images over unsecured channels of communication poses a major risk of revealing
confidential data to unauthorised people[16].
M. T. Elkandoz and W. Alexan identified that developing an image encryption algorithm using chaotic maps was
proven to be most effective because chaotic maps possess some characteristics such as ergodicity and sensitivity
to control parameters and initial conditions. The proposed algorithm by the author starts by shuffling the pixels of
the plain image to create confusion. Then, the shuffled image was diffused by XORing its pixels with a secret
key. This secret key was generated from a combination of different chaotic maps, which were the Arnold cat map,
the Bernoulli map, and the Tent map. Multiple metrics have been used to evaluate the performance of the proposed
scheme. The results proved that the proposed scheme was able to oppose any type of attack, such as visual,
differential, and brute-force attacks. It also shows that the proposed scheme has superior security performance
when compared to some of the other image encryption schemes[17].

464
Tuijin Jishu/Journal of Propulsion Technology
ISSN: 1001-4055
Vol. 45 No. 3 (2024)
__________________________________________________________________________________
A. V Diaconu and K. Loukhaoukha proposed an image encryption scheme that was based on Rubik’s cube
principle. Rubik’s cube principle was used to permute the pixels of the original image. The second step of the
algorithm uses the XOR operation to create confusion between the original and ciphered images. The odd rows
and columns of the image are XORed using a secret key. The same key was rearranged and used for the XORing
of the even rows and columns of the image. Various experimental and statistical tests have been performed to
evaluate the performance of the proposed scheme[18].
A new three-step image encryption algorithm was introduced by W. Alexan et.al. The first step of the algorithm
utilises the Fibonacci sequence, the second step utilises an s-box, and the third step utilises a Tan-Bessel function.
Multiple performance evaluation metrics were used to evaluate the performance of the proposed scheme, such as
MSE, PSNR, and information entropy[19].
S. H. Kamali et.al. developed a new image encryption scheme that is a modified advanced encryption standard
algorithm. The modification made to the AES algorithm was based on shift and row transformations. Shift and
row transformations state that if the values of the first row and column were even, the first and fourth rows remain
unchanged. Then, each byte in the second and third rows was shifted cyclically to the right. However, if the first
and third rows remain unchanged, each byte of the second and fourth rows was shifted to the left. The proposed
scheme has a strong performance against statistical attacks due to its complexity. The encryption time of the
proposed scheme is 8.565 ms, which shows how fast the algorithm is and how it can be used for real-time
encryption[20].
P. Dahiwal and A. Kulkarni proposed the Advanced Encryption Standards (AES) and the XOR technique for
secure image encryption. The author proposed utilising the XOR encryption technique to encrypt segmented
images. The researchers conducted picture transmission experiments employing 3*1, 3*2, and 3*3 images that
were segmented with an encryption method and discovered that the suggested XOR-based image segmentation
surpassed the existing AES approach in terms of time complexity[21].
Huixue Jia et.al. categorized image properties using multilayer homomorphic encryption and image data
partitioning. A novel partitioning method suggested by the author helps to reduces computer complexity and
improves classification accuracy. The author had increased data security and privacy using partitioned image-
specific, homomorphic encryption. The proposed compound encryption method leverages homomorphic
computing to decrease computational and storage overheads and solve encryption's inherent complexity.
Computing efficiency, storage and transmission cost reduction, security, and privacy are better with the proposed
solution. This research provides an innovative and effective technique to image feature classification in cloud
computing and big data[22].
Yuxi Mi et.al presented an in-depth study of the privacy protection of face images. Based on the observations on
model perception and training behaviour, the author presented two methodological advances, pruning low-
frequency components and using randomly selected channels, to address the privacy goal of concealing visual
information and impeding recovery. The author proposed their findings into a novel privacy-preserving face
recognition method, named Partial Face. Extensive experiments demonstrate that Partial Face effectively balances
privacy protection goals and recognition accuracy[23].
To address the privacy preservation of facial images on social media, Xin Dong et.al proposed the frequency-
restricted identity-agnostic (FRIA) framework to encrypt face images with frequency-restricted and identity-
agnostic attacks to address the face privacy leakage problem. FRIA outperforms other state-of-the-art methods in
generating more natural encrypted faces while attaining high black-box attack success rates of 96%. In addition,
Experiments on real-world black-box commercial API further reveal the potential of FRIA in practice[24].
For the face Recognition Integration Platform (FRIP), Sei Nakanishi et al. examined the reaction time of face
recognition engines during authentication using homomorphic encryption. The response time surpasses 3 s with
120 registrants with the current implementation. Maximum cluster size should be under 120. Calculating the
Euclidean distance between encrypted face characteristics took up most of the response time for each step. Parallel
processing with clustering was also studied to minimize reaction time. Five clusters were needed for parallel

465
Tuijin Jishu/Journal of Propulsion Technology
ISSN: 1001-4055
Vol. 45 No. 3 (2024)
__________________________________________________________________________________
processing when the realistic maximum for facial recognition engines was 500. The evaluation showed that
homomorphic encryption provided useful insight into FRIP design[25].
Heping Wen et.al. presented a safe picture encryption technique combining chaos-based block permutation and
weighted bit planes chain diffusion based on a variant structure of classical permutation-diffusion to address the
security issue with image encryption technology. Divide a picture into sub-blocks, block scrambling, block
rotation and inversion, negative-positive transformation, and colour component shuffling are consecutively
conducted with chaotic plaintext association during the permutation phase. The chain diffusion step used various
encryption algorithms for the high and low 4-bit planes based on picture information weight. The method meets
confusion, diffusion, and avalanche cryptographic criteria and has outstanding numerical statistical qualities with
a vast cryptographic space, according to theoretical and empirical assessments[26].
The author unified the task of anonymization and visual identity information hiding and propose a novel face
privacy protection method based on diffusion models, dubbed Diff-Privacy. Specifically, they trained their
proposed multi-scale image inversion module (MSI) to obtain a set of SDM format conditional embeddings of the
original image. Based on the conditional embeddings, they designed corresponding embedding scheduling
strategies and construct different energy functions during the denoising process to achieve anonymization and
visual identity information hiding. Extensive experiments have demonstrated the effectiveness of the proposed
framework in protecting facial privacy[27].
E. Abusham et.al proposed an image encryption scheme to counter spoofing attacks by integrating it into the
pipeline of Linear Discriminant Analysis (LDA) based face recognition. The encryption scheme uses XOR pixels
substitution and cellular automata for scrambling. A single key was used to encrypt the training and testing
datasets in LDA face recognition system. For added security, the encryption step requires input images of faces
to be encrypted with the correct key before the system can recognize the images. An LDA face recognition scheme
based on random forest classifiers has achieved 96.25% accuracy on ORL dataset in classifying encrypted test
face images. In a test where original test face images were not encrypted with keys used for encrypted feature
databases, the proposed system achieved 8.75% accuracy only showing it was capable of resisting spoofing
attacks[28].
Majed Alsafyani et al. developed a revolutionary face feature encryption method that integrates image
optimization, cryptography, and deep learning (DL) architectures. Managing the initial standards of the 5D
conservative chaotic technique using an optical chaotic map improved key security. Facial image encryption and
decryption were completed with a secure Crypto General Adversarial neural network and chaotic optical map.
One "hidden factor" in the machine learning (ML) encryption approach was the target field. A modernizing
network decrypted an image to a unique image. A region-of-interest (ROI) network extracts involved things from
encrypted photos to simplify data mining in a private context. The suggested study shows that the recommended
implementation improves security without compromising image quality. Experimental findings reveal that the
proposed model outperforms existing models in PSNR (92%), RMSE (85%), SSIM (68%), MAP (52%), and
encryption speed (88%)[29].

Fig 3.1. System Architecture

466
Tuijin Jishu/Journal of Propulsion Technology
ISSN: 1001-4055
Vol. 45 No. 3 (2024)
__________________________________________________________________________________
P. Y. Chen et.al used blockchain architecture to create a third-party-free information transfer and communication
system for unmanned stores. The author employed distributed ledgers and encryption/decryption. Google's Face
Net verified retail customers' IDs. Face Net's precision and simplicity make it ideal for unmanned retail
identification verification. A face picture library was needed for facial recognition. For encryption and decryption,
author had utilized RSA encryption technique. P2P networks broadcasted messages using a directed acyclic graph
to avoid endless cycles. The system uses a sidechain as its major endpoint design to benefit from a centralized
network while retaining ledger impartiality regulated by lower-authority users to avoid transaction record
tampering[30].
Xiao Yang et.al developed a technique that can encrypt the personal photos such that they can protect users from
unauthorized face recognition systems but remain visually identical to the original version for human beings. To
achieve this, the author(s) proposed a targeted identity-protection iterative method (TIP-IM) to generate
adversarial identity masks which can be overlaid on facial images, such that the original identities can be
concealed without sacrificing the visual quality. Extensive experiments demonstrated that TIP-IM provides 95%+
protection success rate against various state-of-the-art face recognition models under practical test scenarios.
Besides, they also showed the practical and effective applicability of the proposed method on a commercial API
service[31].
Luoyin Feng et.al proposed an image recognition and encryption algorithm based on an artificial neural network
and multidimensional chaotic sequence. It adopts the combination of high-dimensional chaos and one-
dimensional chaos to enhance the security of the key. And, complete the image encryption in the DWT-DCT
transform domain and use PCA and BP neural network to achieve face recognition. Then, the key sensitivity,
algorithm recognition rate, and robustness of the encryption algorithm were analyzed and tested. Finally, the
algorithm recognition rate and robustness of the three encrypted face recognition algorithms and unencrypted face
recognition were compared and analyzed. The results show that the recognition rate of the face recognition method
using transform domain encryption was not much different from that of the unencrypted face recognition method,
and it has good robustness. And, after encryption, the security of face image data on the Internet was greatly
improved[32].
3. Methodologies
This section describes the methodologies of image encryption and compression techniques for the face recognition
system used in this research work. Figure 3.1 describes the detailed system architecture of image encryption and
transmission over cloud system.
The outlined process describes the methodology for image encryption and transmission in the context of a face
recognition system, specifically utilizing Amazon Web Services (AWS) for cloud storage. Here's a breakdown of
the steps involved:
A. Face Image Capture: The process begins with capturing the face image using the camera attached to
the user's desktop or laptop. This step involves acquiring the raw image data of the user's face.
B. Image Segmentation: The original face image is segmented into different combinations of grids, such
as 3x1, 3x2, and 3x3. This segmentation process divides the face image into smaller sections or blocks.
C. Encryption Technique Application: Following segmentation, an encryption technique, specifically
XOR encryption operations, is applied to each segmented image block. XOR encryption involves applying the
XOR operation between the segmented image data and a cryptographic key. This step aims to protect the
confidentiality of the facial features by scrambling the pixel values.
D. Cloud Upload: Once encrypted, the segmented images are uploaded to the cloud storage service
provided by Amazon Web Services (AWS). AWS offers various storage options, such as Amazon S3 (Simple
Storage Service), which provide scalable and secure storage for digital assets like images.

467
Tuijin Jishu/Journal of Propulsion Technology
ISSN: 1001-4055
Vol. 45 No. 3 (2024)
__________________________________________________________________________________
E. Storage and Management: Within the AWS cloud infrastructure, the encrypted segmented images are
stored securely. Access controls and permissions can be applied to restrict access to authorized users or systems,
ensuring the confidentiality and integrity of the stored images.
F. Access and Retrieval: Authorized users or systems can access the encrypted images stored in the AWS
cloud for further processing or recognition tasks. Access to the stored images is governed by authentication
mechanisms and access control policies implemented within the AWS environment.
G. Decryption and Recognition: When required, the encrypted segmented images can be retrieved from
the AWS cloud, decrypted using the appropriate decryption keys, and then processed for face recognition tasks.
Decryption reverses the encryption process, allowing the original facial features to be reconstructed and used for
recognition purposes.
By following this methodology, the face recognition system can ensure the secure transmission and storage of
facial images on the AWS cloud, leveraging encryption techniques to protect sensitive data while facilitating
efficient access and retrieval for recognition tasks.
3.1 XOR Encryption and Decryption
The XOR cipher is indeed a simple yet powerful cryptographic method that relies on the bitwise exclusive OR
operation. It is commonly used due to its simplicity and computational efficiency. Here are some key principles
and characteristics of the XOR cipher:
A. Encryption and Decryption:
a. To encrypt a plaintext message M, the message is XORed with a secret key K to produce the encrypted
message E, denoted as M⊕K=E.
b. Similarly, to decrypt the encrypted message E, it is XORed again with the same secret key K, resulting in the
original plaintext message M, expressed as E⊕K=M.
c. This makes XOR cipher a symmetric encryption algorithm, where the same key is used for both encryption
and decryption, simplifying the implementation.
B. Key Length and Security:
a. If the key is shorter than the message and repeated, the encryption is vulnerable to frequency analysis, where
patterns in the plaintext become apparent in the ciphertext.
b. However, if the key is as long as the message and used only once, breaking the encryption becomes impractical
due to the sheer number of possible keys (2^N for an N-bit key).
c. Sharing a long key securely can be challenging, but using a pseudo-random number generator (RNG) to
produce a stream of keys can mitigate this issue, requiring only the initial seed to be shared.
C. Properties of XOR Operation:
a. A⊕0=A: XORing any value with 0 results in the original value.
b. A⊕A=0: XORing any value with itself results in 0.
c. (A⊕B) ⊕C=A if and only if B⊕C=0: XORing A with the result of XORing B and C is equivalent to XORing
A with B.
d. (B⊕A) ⊕A=B if and only if B⊕A=B: XORing B with the result of XORing A and A is equivalent to XORing
B with 0, resulting in B.
D. XOR Algorithm:
Algorithm: EncryptMessage (M, K)

468
Tuijin Jishu/Journal of Propulsion Technology
ISSN: 1001-4055
Vol. 45 No. 3 (2024)
__________________________________________________________________________________
Input:
M: Plaintext message
K: Secret key
Output:
E: Encrypted message

1. Initialize E as an empty string

2. key_length ← length of K
3. For i from 0 to length of M - 1 do
a. plaintext_char ← ASCII value of M[i]
b. key_char ← ASCII value of K[i % key_length]
c. encrypted_char ← plaintext_char XOR key_char
d. Append character representation of encrypted_char to E
4. Return E

4. Results and Discussion

The results were drawn by segmentation of the image into different matrices like 3*1,3*2 and 3*3 which means
the image gets segmented into 3,6 & 9 segments. After segmentation, the XOR Encryption algorithms are applied
to images for secure transmission over cloud as described in section 3.0
4.1 Experimental Setup:
a) All measurements and experiments are conducted on a single core of an Intel Core i5 CPU running at
2600 MHz, with 8 GB RAM.
b) The experiments are carried out using the NETBEANS 8.2 Integrated Development Environment
(IDE), with Java as the programming language. This implies that the encryption and segmentation algorithms are
implemented using Java programming language within the NETBEANS environment.
We used the AWS cloud to upload the facial images for encryption using XOR. Our findings are summarized in
Tables I for XOR encryption mechanisms.
Table I: XOR Encryption

Initial Image Size Final Image Size Execution Time

XOR
(KB) (KB) (ms)
3*1 51.3 52.7 4362
3*2 52.8 56.2 5114
3*3 52.4 61.7 5161

469
Tuijin Jishu/Journal of Propulsion Technology
ISSN: 1001-4055
Vol. 45 No. 3 (2024)
__________________________________________________________________________________

Fig 4.1. XOR Encryption

From analysis of Table I and Figure 4.1, it shows that the original file was encrypted using XOR operations with
3*1,3*2,3*3 Segmentation and the encrypted file size is slightly bigger in size compared to the original size.
Table II: Execution Time

Segmentation XOR-Execution Time(ms)

3*1 55
3*2 60
3*3 75

In Table II, we show the execution time required for XOR image encryption technique for transmission of 3
segmented images on cloud platform. It is observed that in all 3 segmentation matrices, the XOR image encryption
mechanism has less time needed to encrypt the image. In our performance analysis, the most important point is
the time complexity for secured image authentication on the cloud, in that regard the XOR encryption considerably
faster. Figure 4.2 shows that the XOR image encryption time ranges from just 55 ms to 75 ms for 3*1,3*2 and
3*3 image segmentation technique respectively. The proposed algorithm of XOR with image segmentation has
been proven to be way faster and efficient for image transmission.

Fig 4.2. Transmission Time over Cloud

5. Conclusion
The research presents efficient methodologies for ensuring secure transmission of images across cloud networks
through the use of compression, encryption, and image segmentation algorithms. Specifically, XOR image
encryption algorithms are proposed and implemented in the study. The key highlights of the research findings are
as follows:

470
Tuijin Jishu/Journal of Propulsion Technology
ISSN: 1001-4055
Vol. 45 No. 3 (2024)
__________________________________________________________________________________
Image Segmentation: The images are initially divided into matrices of sizes 3x1, 3x2, and 3x3. This segmentation
approach breaks down the images into smaller parts, facilitating efficient processing and transmission.
XOR Encryption: The XOR encryption algorithm is applied to encrypt the segmented images. This
cryptographic technique ensures the confidentiality and integrity of the image data during transmission over the
cloud network.
Exceptional Speed: The research findings indicate that the proposed XOR encryption technology, combined with
image segmentation, achieves exceptional speed in transmitting data securely via cloud networks. The time taken
for encryption falls within the range of 50-65 milliseconds, demonstrating high efficiency.
Implications for Cloud Authentication: The rapid transmission speed of encrypted images suggests potential
applications in cloud-based authentication systems. The quick encryption process enables prompt authentication,
which is crucial in various industries where facial cognitive authentication techniques are in demand.
Overall, the research contributes to the advancement of secure image transmission methodologies in cloud
environments. The combination of image segmentation and XOR encryption offers a practical solution for
ensuring data security while maintaining efficient transmission speeds. The findings hold promise for enhancing
authentication processes and meeting the needs of industries relying on facial cognitive authentication techniques.
References
[1] X. Wang, Z. Yan, R. Zhang, and P. Zhang, “Attacks and defenses in user authentication systems: A survey,”
J. Netw. Comput. Appl., vol. 188, p. 103080, Aug. 2021, doi: 10.1016/J.JNCA.2021.103080.
[2] O. MuhtahirO., A. Adeyinka O., and A. Kayode S., “Fingerprint Biometric Authentication for Enhancing
Staff Attendance System,” Int. J. Appl. Inf. Syst., vol. 5, no. 3, pp. 19–24, Feb. 2013, doi: 10.5120/IJAIS12-
450867.
[3] B. Zhou, Z. Xie, and F. Ye, “Multi-Modal Face Authentication using Deep Visual and Acoustic Features,”
IEEE Int. Conf. Commun., vol. 2019-May, May 2019, doi: 10.1109/ICC.2019.8761776.
[4] D. R. Kisku and R. D. Rakshit, “Face Spoofing and Counter-Spoofing: A Survey of State-of-the-art
Algorithms,” Trans. Eng. Comput. Sci., vol. 5, no. 2, pp. 31–31, May 2017, doi: 10.14738/TMLAI.52.3130.
[5] A. Adler and S. A. C. Schuckers, “Biometric Vulnerabilities, Overview,” Encycl. Biometrics, pp. 271–279,
2015, doi: 10.1007/978-1-4899-7488-4_65.
[6] P. Nagarsheth, E. Khoury, K. Patil, and M. Garland, “Replay Attack Detection Using DNN for Channel
Discrimination,” Proc. Annu. Conf. Int. Speech Commun. Assoc. INTERSPEECH, vol. 2017-August, pp.
97–101, 2017, doi: 10.21437/INTERSPEECH.2017-1377.
[7] M. Witkowski, S. Kacprzak, P. Zelasko, K. Kowalczyk, and J. Gałka, “Audio Replay Attack Detection Using
High-Frequency Features,” Proc. Annu. Conf. Int. Speech Commun. Assoc. INTERSPEECH, vol. 2017-
August, pp. 27–31, 2017, doi: 10.21437/INTERSPEECH.2017-776.
[8] R. M. Redlich and M. A. Nemzow, Digital information infrastructure and method for security designated
data and with granular data stores: US, US9734169 [P]. 2017.
[9] Z. Han, S. Huang, H. Li, and N. Ren, “Risk assessment of digital library information security: A case study,”
Electron. Libr., vol. 34, pp. 471–487, Jun. 2016, doi: 10.1108/EL-09-2014-0158.
[10] C. Zhou, Y. Guo, W. Huang, H. Jiang, B. Li, and J. Chen, “Information security defense method of electric
power control system based on digital watermark,” no. Icmemtc, pp. 174–179, 2016, doi: 10.2991/icmemtc-
16.2016.32.
[11] E. Chisanga and E. K. Ngassam, “Towards a conceptual framework for information security digital divide,”
2017 IST-Africa Week Conf. IST-Africa 2017, Nov. 2017, doi: 10.23919/ISTAFRICA.2017.8102398.
[12] M. T. Elkandoz, W. Alexan, and H. H. Hussein, “Logistic Sine Map Based Image Encryption,” Signal
Process. - Algorithms, Archit. Arrange. Appl. Conf. Proceedings, SPA, vol. 2019-September, pp. 290–295,
Sep. 2019, doi: 10.23919/SPA.2019.8936718.
[13] I. Younas and M. Khan, “A New Efficient Digital Image Encryption Based on Inverse Left Almost Semi
Group and Lorenz Chaotic System,” Entropy 2018, Vol. 20, Page 913, vol. 20, no. 12, p. 913, Nov. 2018,
doi: 10.3390/E20120913.

471
Tuijin Jishu/Journal of Propulsion Technology
ISSN: 1001-4055
Vol. 45 No. 3 (2024)
__________________________________________________________________________________
[14] S. Kanwal, S. Inam, O. Cheikhrouhou, K. Mahnoor, A. Zaguia, and H. Hamam, “Analytic Study of a Novel
Color Image Encryption Method Based on the Chaos System and Color Codes,” Complexity, vol. 2021,
2021, doi: 10.1155/2021/5499538.
[15] F. Khan, J. S. Khan, J. Ahmad, and M. Khattak, “A novel image encryption based on Lorenz equation,
Gingerbreadman chaotic map and S8 permutation,” J. Intell. Fuzzy Syst., vol. 33, pp. 1–13, Oct. 2017, doi:
10.3233/JIFS-17656.
[16] A. Sinha and K. Singh, “A technique for image encryption using digital signature,” Opt. Commun., vol. 218,
no. 4–6, pp. 229–234, Apr. 2003, doi: 10.1016/S0030-4018(03)01261-6.
[17] M. T. Elkandoz and W. Alexan, “Image encryption based on a combination of multiple chaotic maps,”
Multimed. Tools Appl., vol. 81, no. 18, pp. 25497–25518, Jul. 2022, doi: 10.1007/S11042-022-12595-
8/METRICS.
[18] A. V Diaconu and K. Loukhaoukha, “An improved secure image encryption algorithm based on Rubik’s
cube principle and digital chaotic cipher [J],” Math. Probl. Eng., vol. 2013, 2013, doi: 10.1155/2013/848392.
[19] W. Alexan, M. Elbeltagy, A. Aboshousha, and H. Hussein, Image Encryption Through Fibonacci Sequence,
S-Box, and Tan Bessel Function. 2021.
[20] S. H. Kamali, M. Hedayati, R. Shakerian, and M. Rahmani, “A new modified version of Advanced
Encryption Standard based algorithm for image encryption,” ICEIE 2010 - 2010 Int. Conf. Electron. Inf.
Eng. Proc., vol. 1, 2010, doi: 10.1109/ICEIE.2010.5559902.
[21] P. Dahiwal and A. Kulkarni, “Facial Image Encryption & Compression for Secure Image Transmission on
Cloud Storage,” Int. J. Intell. Syst. Appl. Eng., vol. 12, no. 6s, pp. 615–621, Nov. 2023, Accessed: Mar. 08,
2024. [Online]. Available: https://ijisae.org/index.php/IJISAE/article/view/4000
[22] H. Jia et al., “Efficient and privacy-preserving image classification using homomorphic encryption and
chunk-based convolutional neural network,” J. Cloud Comput., vol. 12, no. 1, 2023, doi: 10.1186/s13677-
023-00537-0.
[23] Y. Mi et al., “Privacy-Preserving Face Recognition Using Random Frequency Components,” Proc. IEEE
Int. Conf. Comput. Vis., pp. 19616–19627, 2023, doi: 10.1109/ICCV51070.2023.01802.
[24] X. Dong, R. Wang, S. Liang, A. Liu, and L. Jing, “Face Encryption via Frequency-Restricted Identity-
Agnostic Attacks,” MM 2023 - Proc. 31st ACM Int. Conf. Multimed., pp. 579–588, 2023, doi:
10.1145/3581783.3612233.
[25] S. Nakanishi, Y. Narusue, and H. Morikawa, “Response Time of Cloud-Based Facial Recognition System
Utilizing Homomorphic Encryption,” IEICE Commun. Express, vol. 12, no. 12, pp. 603–606, 2023, doi:
10.23919/comex.2023col0010.
[26] H. Wen, Y. Lin, S. Kang, X. Zhang, and K. Zou, “Secure image encryption algorithm using chaos-based
block permutation and weighted bit planes chain diffusion,” iScience, vol. 27, no. 1, p. 108610, 2024, doi:
10.1016/j.isci.2023.108610.
[27] D. For, “T Wilight S Hadows on the B Right F Ace P Aul D Ale,” pp. 1–20, 1996.
[28] E. Abusham, B. Ibrahim, K. Zia, and M. Rehman, “Facial Image Encryption for Secure Face Recognition
System,” Electron., vol. 12, no. 3, 2023, doi: 10.3390/electronics12030774.
[29] M. Alsafyani, F. Alhomayani, H. Alsuwat, and E. Alsuwat, “Face Image Encryption Based on Feature with
Optimization Using Secure Crypto General Adversarial Neural Network and Optical Chaotic Map,” Sensors,
vol. 23, no. 3, 2023, doi: 10.3390/s23031415.
[30] P. Y. Chen, Y. C. Cheng, N. S. Pai, and Y. H. Chiang, “Applying Blockchain Technology and Facial
Recognition to Unmanned Stores,” Sensors Mater., vol. 35, no. 6, pp. 2081–2100, 2023, doi:
10.18494/SAM4289.
[31] X. Yang et al., “Towards Face Encryption by Generating Adversarial Identity Masks,” Proc. IEEE Int. Conf.
Comput. Vis., pp. 3877–3887, 2021, doi: 10.1109/ICCV48922.2021.00387.
[32] L. Feng and X. Chen, “Image Recognition and Encryption Algorithm Based on Artificial Neural Network
and Multidimensional Chaotic Sequence,” Comput. Intell. Neurosci., vol. 2022, 2022, doi:
10.1155/2022/9576184

472