Securing Visual Data: An R-Channel LSB

for Image Stegnography

Mrs. Shylaja D N, Suvam Sharma, Shreya Kumari, Saurabh Sharma Timilsina, Rishav Upadhyay
Department of Computer Science and Engineering
Bangalore Technological Institute, Bengaluru, India
{[email protected], [email protected], [email protected], [email protected],
[email protected]}

Abstract—In the 21st century, the challenge of securely

transmitting information over digital mediums persists due to the
Image steganography serves as a protective measure for
looming threat of interception by unauthorized entities. sensitive information by concealing it within images, thereby
Steganography emerges as a pivotal solution, enabling the evading detection by the human sense of vision. Its
concealment of sensitive data within seemingly innocuous files, applications span diverse fields including communications,
thereby bolstering security during communication. This research mobile computing, online voting systems, surveillance, and
paper presents an investigation into the efficacy of various least medical applications for securing medical records [9]-[10].
significant bit (LSB) variants within steganographic techniques. A Over the years, numerous image steganography techniques
comparative analysis is conducted across three prevalent image have been developed and implemented to enhance data
formats—PNG, Bitmap, and GIF—assessing performance through
security. Among these techniques, the Least Significant Bit
metrics such as mean square error (MSE) and peak signal-to-noise
ratio (PSNR). The results reveal that PNG and Bitmap formats (LSB) method stands out as one of the simplest methods
exhibit superior performance in concealing data while upholding falling under the spatial domain category. This method
image quality compared to GIF. These findings offer valuable involves encoding confidential information into the least
insights for practitioners and researchers seeking to optimize significant bits of each pixel within an image. By doing so, the
steganographic methods for secure information transmission. LSB technique provides a certain level of confidentiality when
transmitting information over the internet. One of the key
Keywords— Stegobit, Steganography, Stego-image, LSB, PSNR, advantages of the LSB method lies in its ability to achieve low
MSR values of Mean Square Error (MSE) and high values of Peak
I. INTRODUCTION Signal-to-Noise Ratio (PSNR). These characteristics make it
challenging for the human eye to discern any alterations made
to the image, thereby ensuring a level of stealthiness in the
Recent advancements in multimedia technologies have concealment process. Due to these advantages, the LSB
significantly enhanced the efficiency and speed of information method is widely utilized in the development of image
exchange over networks. The global proliferation of the internet, steganography applications, offering a reliable means of
driven by digitalization, has further facilitated data sharing and securing sensitive information during transmission.
utilization through cloud-based infrastructures [1]-[4]. However,
safeguarding information from unauthorized access poses a A K-LSB-based method was implemented for concealing one
formidable challenge in modern societies. To address this image within another image by utilizing the K-least significant
concern, cryptography and steganography techniques have bits in [11]. In [12], An enhanced LSB substitution technique
emerged as pivotal mechanisms for ensuring data security [5]- for image-based steganography was implemented, leveraging a
[6]. local entropy filter for identifying the bit containing the
concealed image. This approach represents a significant
Cryptographic techniques play a crucial role in securing advancement in the field, as it allows for the concealment of
information by encrypting it, thereby rearranging it in a images with minimal distortion and loss of information. The
manner that makes it difficult for intruders to detect and technique employed in this study aimed to bolster security
intercept. Basic authentication methods encompass hashing, measures by incorporating a two-step process. Initially, the
digital signatures, and establishing threshold values for message intended for embedding was encrypted, thereby
embedding secret bits. However, cryptography alone may not adding an extra layer of security. Subsequently, the encrypted
always provide sufficient assurance that confidential message was converted into a stego-image using the
information is entirely protected. Thus, there has been a implemented method. This approach not only enhances data
growing recognition of the need to complement cryptography security but also results in a notable increase in embedding
with steganography [7]-[8]. Indeed, experts commonly capacity while maintaining a high Peak Signal-to-Noise Ratio
advocates for the integration of both steganography and (PSNR).
cryptograhy as multiple layers of security. This multilayered
approach enhances the robustness of data protection measures, In [13], an automated method for securing a secret message
offering added resilience against unauthorized access and was developed, incorporating a dual level of security. Initially,
inception. the secret message underwent encryption using a Java-based
encryption method known as Character Bit Shuffler (CBS),
constituting the first level of security. Subsequently, the
encrypted secret message bits were embedded within an
image utilizing the least significant bit (LSB) technique,
thereby establishing the second level of security. The
adopted method was chosen for its simplicity and its
capacity to preserve the quality of the image. In [14], an
enhanced LSB Steganography approach was introduced,
leveraging the modulus function for embedding data
concealment information. In [15], a method was devised for
concealing secret message bits within the pixels of an RGB
true-color image, utilizing a secret key for enhanced
security. Moreover, the transmission capacity of the image
was augmented by compressing the secret data, originally
consisting of seven bits in ASCII format, into five-bit
streams.

This enables the recovery of the original secret messages

Fig 1: A Typical Cryptographic System
embedded by the sender. The depicted processes elucidate the
fundamental operations involved in steganography, facilitating
secure communication and data concealment between parties. B. The mathematical formulation for the LSB Stenography
Techniques
The study's findings indicate a notable enhancement in the In this study, the LSB substitution method was employed to
image capacity, with an increase of up to 30% achieved embed text within 8-bit grayscale cover images (GIF) as well as
through the implemented LSB substitution variants. 24-bit color images (Bitmap and PNG).
Additionally, the Peak Signal-to-Noise Ratio (PSNR)
For an 8-bit grayscale image denoted as Igrayscale of size, PI gray × QI
demonstrated improvement by up to 27% compared to
gray, each pixel can be represented using equation 1 as follows:
existing techniques. Hence, the primary objective of this study
is to implement various variants of the LSB substitution
I gray =
method, denoted as Stego 1, 2, 3, and 4, across three distinct

{x }
image formats: GIF, Bitmap, and PNG. Subsequently, a
comprehensive comparative analysis will be conducted to ij | 0 ≤ i ≤ PI gray, 0 ≤ i ≤ QI gray, x ij ∈ {0,1,2,…,255} (1)
evaluate the performance of these variants.

II. METHODOLOGY For a 24-bit color cover image denoted as Icolor of size PIcolor ×
A. Overview of the images-based steganography QIcolor is given in equation 2 as follows:
Fig.1 provides a visual representation of the process
involved in concealing secret messages within a given Icolor red =
medium, along with an overview of the components
comprising a typical steganographic system. The
illustration delineates two primary processes: embedding X red |0 ≤ i ≤ PIcolor red , xred ∈{0,1,2.....,255} (2)
ij
and extraction of text-based messages, carried out at the ij |0 ≤ i ≤ QIcolor red ,
sender and receiver's ends, respectively.
During the embedding process, secret messages are The n-bit of the secret message to be embedded into the 8-bit
concealed within a cover image using a stego-key, resulting grayscale and 24-bit color image, denoted as S, is defined by
in the generation of a stego-image. Subsequently, the stego- equation 3.
image is transmitted over a communication medium to the
S = {si | 0 ≤ i ≤ n, si ∈{0,1}}
intended recipient. At the receiver's end, the extraction
process involves retrieving the concealed messages from (3)
the received stego-image using the corresponding stego-
key. This enables the recovery of the original secret
messages embedded by the sender. The depicted processes
Furthermore, the secret message is rearranged to form a k-bit
elucidate the fundamental operations involved in
virtual message S ′ , which is defined in equation 4 as follows:
steganography, facilitating secure communication and data
concealment between parties.
At the receiver's end, the secret message is extracted from
the stego-image using the provided stego-key. Once the
'
{ ' '
S = s i|0 ≤ i ≤ n , si ∈ 0,1, .. , 2 { k −1
}}
extraction process is completed successfully, the concealed
message is revealed in its original text format. This final
(4)
step allows the recipient to access and interpret the hidden
'
information, ensuring the secure and accurate transmission Where n = PI gray× QI gray and n = PI color× QI color
of the intended message.
Furthermore, a mapping between the original secret message (S) III. RESULTS AND DISCUSSION
and the embedded message (S ′) is established, yielding equation
5 as follows: A. Computation of peak signal to noise ratio (PSNR)
k=1
Figure 2 illustrates a bar graph representing the computed average
'
s i = ∑ si × k + j × 2
k1 j
(5) PSNR for the considered variants (Stego 1, 2, 3, and 4), tested with
j=0 PNG, BMP, and GIF images. The graph demonstrates that as the
number of stego bits used to embed the secret message increases,
Therefore, the embedding process is completed for each pixel, the average PSNR value decreases. This observation indicates that
with k (where, k=1,2,3,4) bits of the rearranged message S ′ as more stego bits are replaced by the message bits, the stego-
stored in each pixel. Thus, each pixel stores the k-bit message image becomes more distorted, posing a potential security threat to
for a stego-pixel, which is formulated by equation 6 as follows: the concealed message.
Furthermore, the graph reveals that the PSNR for PNG images is
Xi' = xi mod2k + si (6) relatively higher for stego 1-3 bits compared to GIF images. The
PSNR for GIF images is relatively low compared to other image
formats, suggesting that GIF images may experience more
Lastly, the embedded image is extracted from the
distortion when used for steganographic embedding. These
stego- image using equation 7.
findings underscore the importance of selecting appropriate stego
parameters and image formats to ensure optimal security and
s'= xi mod 2k (7) image quality in steganographic applications.

C. Modalities of Data Collection and Criteria for

Evaluation
For the evaluation of the technique, ten images were
obtained for each of the formats, namely GIF, Bitmap,
and PNG, sourced from the internet. These images
were utilized to assess the performance of the embed-
ding process and its impact on image quality. By ana-
lyzing PSNR and MSE values for the stego-images
across the different image formats, the effectiveness of
the embedding technique can be assessed, providing
insights into its performance and suitability for Figure 2: Bar chart showing the PSNR computation
concealing information within digital images.
B. Analysis for mean square error (MSE) and stego bits
Mean Square Error (MSE) provides an estimate of the
resultant error in the cover image due to the embedded The line plot in Fig. 3 depicts the computed average MSE values
data. It can be computed mathematically using Equa- across the considered image formats for stego 1-4 bits. It was
tion 8 as follows: observed that the error for the three image formats was very
minimal at stego 1 bit. However, as the number of replacing stego
bits increases, the errors become more significant. It was noted that
N the errors in GIF images become very significant compared to
MSE = ∑ i (Ci - C ′)2 (8) other image formats (PNG and Bitmap) as the number of stego bits
i=1
increases. This observation suggests that GIF images may be more
susceptible to distortion when used for steganographic embedding,
Where Ci denotes the size of the stego-image, CiI is the size of especially with higher levels of data concealment. These findings
the cover image and N is the size of the original image. emphasize the importance of considering both the stego parameters
PSNR, or Peak Signal-to-Noise Ratio, is indeed utilized to and the characteristics of the image format when implementing
quantify the degree of distortion introduced by the embedded steganographic techniques to ensure minimal distortion and
data. The mathematical formula for calculating PSNR is given optimal security.
by equation 9 as follows:
PSNR(decibel) = 10 log10 (255/MSE) (9)

A high computed value of PSNR indeed indicates that the

image has undergone less distortion due to the embedded data,
suggesting higher image fidelity and better quality.
Conversely, a lower PSNR value suggests that the image has
experienced more distortion, indicating poorer image quality. Figure 3: MSE computation across all image formats
 CONCLUSION
This study has conducted a
comparative analysis on the
performance of the LSB substitution
method across three different image
formats. Throughout the evaluation
process, it was observed that as the
number of replacing stego bits
increases, there is a consistent relative
drop in the PSNR values, accompanied
by an increase in MSE. These trends
indicate a progressive distortion of the
images used for hiding the secret
message.
It's important to recognize that the
security of the embedded message
largely relies on the perceptibility of
the image. Since steganographic
methods are not completely foolproof,
it is imperative to exercise caution
when concealing sensitive information.
While steganography can offer a layer
of security for transmitting covert
messages, it's essential to assess the
trade-offs between security and
perceptibility, and to implement
additional encryption or security
measures when dealing with highly
sensitive data. Ultimately, a balanced
approach to data concealment is crucial
to ensuring both security and usability
in steganographic applications.
Based on the results obtained from the
study, it is recommended that the cover
picture used for steganographic
embedding should have sufficient
capacity to accommodate the secret
data. Specifically, the secret data
should occupy no more than 20% of
the cover image on average. This
ensures that the embedding process
does not excessively distort the cover
image, thereby maintaining its
perceptual quality and reducing the risk
of detection.
Additionally, it is advisable to
minimize the amount of data to be
hidden within the cover image. By
keeping the data payload small, the
number of pixels required for
embedding can be minimized,
mitigating the potential for noticeable
distortion in the stego-image. This
approach helps to strike a balance
between embedding capacity and
image quality, ensuring that the
concealed information remains covert
while preserving the integrity of the
cover image.
D. Review of the Development Stack B. Assumptions
1) React: React is a JavaScript library for building user Developing a ride-hailing dApp using JavaScript for the
interfaces. It allows developers to create interactive UI frontend and Solidity for smart contracts presents a unique set
components efficiently, making it an ideal choice for of challenges. While libraries exist to streamline dApp
developing frontend applications. Our decision to use React creation, their immaturity might necessitate significant custom
was influenced by its popularity and robust ecosystem, programming, particularly at the API level where
enabling seamless integration with other technologies in our functionalities should ideally be readily available.
stack.
Furthermore, the differing data types between Solidity and
2) Solidity: It is a high-level language specifically designed for
JavaScript require meticulous development practices to
writing smart contracts on blockchain platforms like
prevent unintended type conversions that could introduce
Ethereum. With its object-oriented structure and support for
unpredictable behaviour within the dApp. Another hurdle
complex data types, Solidity simplifies the development of
arises from MongoDB's eventual consistency model. This
decentralized applications (dApps) by providing a secure and
means there might be temporary inconsistencies between the
efficient way to implement smart contract logic.
driver and rider data stored in the database and the information
3) MongoDB: It is a widely used NoSQL database that offers reflected in the smart contracts. While this doesn't cause
flexibility and scalability for storing and managing data. By logical errors in the ride-hailing process itself, it could impact
adopting MongoDB in our development stack, we aim to users with slower devices. Finally, unlike desktop applications
leverage its document-oriented model to efficiently handle the that can run their own MongoDB instance, web apps typically
storage and retrieval of user and transactional data within our cannot due to performance considerations. To maintain a
decentralized ride-booking application. smooth user experience, the web app will likely rely on an
4) Alchemy: It is a blockchain infrastructure platform that external server running the MongoDB database. This
provides developers with tools and services for building and introduces a degree of centralization, but it's a necessary trade-
scaling decentralized applications. With features like reliable off for ensuring optimal performance and keeping operational
node infrastructure, real-time event streaming, and analytics, costs under control.
Alchemy simplifies the integration of blockchain functionality In essence, building a ride-hailing dApp with this architecture
into our application, ensuring optimal performance and requires careful planning to address data type mismatches,
reliability. potential data inconsistencies due to eventual consistency, and
5) Socket.IO: It is a JavaScript library that enables real-time, the inherent centralization that comes with using an external
bidirectional communication between web clients and servers. MongoDB server. By strategically mitigating these challenges,
By implementing Socket.IO in our application, we can developers can create a dApp that offers a seamless user
facilitate seamless communication between drivers and riders experience for both riders and drivers.
in real- time, enhancing the overall user experience and
efficiency of our decentralized ride-booking platform.
C. Implementation
1) Signup System: The signup system will be included for both
III. METHODOLOGY AND APPROACH the rider and driver side of the application. It is relatively well
designed as it is assumed both the rider and driver have
A. Scope Ethereum wallets (Meta Mask). Therefore, their private and
The web application will be built for any responsive devices public keys can be used to uniquely identify the entities. The
and involves interfaces keeping in mind two kinds of users: signup process will involve typical signup pages for the users
the rider and the driver. The primary aim of theapplication is and their detailswill be stored on the MongoDB. The hash for
to pair the riders who want to go to a destinationwith drivers these details will then be stored in the deployed smart contract
who are willing to take them there. Login and registration which exists on the Ethereum blockchain.
features will have to be built for each type of entity. For the
2) Login System: The login system will also involve typical
rider’s side specifically, pages for being able to select start and
user interfaces for both the driver and rider. In both cases, for
end destinations which provide map view of both, pages for
the login and the signup, the private key of the user will act as
viewing profiles and shortlisting cabs from list of prospective
secrets like the password which they will have to maintain.
types of vehicles and pages for facilitating the payment to
The login system will look up their details on the MongoDB
thedriver and rating them at the end of the ride will have to be
database
programmed. For the driver’s side, pages for confirming ride
requests from riders, map view of start and end locations and 3) Driver and Rider Pairing: This is the crux of the system.
payment screens along with rating screens will have to be The details of the driver are intermittently updated and polled
programmed. through Socket.io. This means when the rider requests for a
ride they get details as new as efficiently possible. The
prospective driver details are generated and showed on the
rider’s dashboard.
4) Ethereum transfer: After the ride is completed, an easy endpoints and route, pages for rating the driver and the rider and
interface is created to transfer the Ethereum tokens from the pages for making the final transaction.
rider’s crypto wallet (MetaMask) to the driver’s crypto wallet
automatically.The cost is calculated by a function that depends
on the distanceof travel and the time it took for the journey to B. Blockchain
complete; distance of travel is calculated through Distance We are using the Ethereum blockchain to deploy our smart
Matrix Algorithm which is embedded within Google Maps contracts. The smart contracts themselves are written in the
API. solidity programming language which is an object-oriented
5) Effortless device responsive system: Our dApp prioritizes a language for defining the structure and methods involved in an
smooth ride, not just on the road! We ensure a seamless application. Our application involves methods for starting a
experience across all devices (mobile, tablet, desktop) with a ride, ending a ride, initializing a ride, pairing a rider to a driver
responsive design that adapts perfectly to your screen. No by making a request through socket.io, et cetera.
more zooming and scrolling, just a perfectly sized interface for
a stress-free ride-hailing experience.

Fig 1: Architecture

D. Benefits of Proposed Implementation

C. No-SQL DBMS
Crypto-Commute streamlines user experience and operations.
Our web app eliminates the need for ride-hailing middlemen, We've chosen MongoDB as our database solution for storing
reducing costs and fostering direct interaction. The user- driver and rider information because of its flexibility and
friendly interface requires no special skills beyond basic web scalability. Unlike traditional relational databases, MongoDB
browsing and an Ethereum wallet (easily set up on uses a schema-less design, which means it can efficiently store
smartphones). No specialized hardware is needed, as Crypto- the varied data structures associated with drivers and riders.
Commute runs on standard devices. Cross-platform This includes details like names, locations, vehicle
compatibility ensures broad accessibility, and the web-based information (for drivers), and preferences. Additionally,
design allows for easy integration with existing technologies MongoDB's horizontal scaling capabilities allow us to easily
and future scalability, positioning Crypto-Commute for accommodate a growing user base without compromising
seamless growth within the ride-hailing landscape. performance. Since transaction details are secured on the
blockchain, MongoDB focuses on the core user data that
IV. SYSTEM ARCHITECTURE facilitates seamless connections between riders and drivers
within the dApp.
Fig. 1 portrays a graphical representation of the system
architecture. Its principal components are as follows:
V. IMPLEMENTATION
A. Frontend
The finally developed system has the following flow of
The front end is built on top of the React framework using
execution:
JavaScript and TypeScript programming languages. It includes
1. The User logs in or signs into the application. Their
pages to signup, login for both the rider and the driver.
details are stored in the MongoDB or the user can sign in
Components to select prospective type of vehicle, pages to
through MetaMask if not already present. The immutable
accept or reject requesting riders, component to select journey
hash of this data is stored on the smart contract.
2. The rider enters details of the start and end position of their
ride. These are used to semi-initialize a new block on the
smart contract. These details are then sent to the relay
server
3. The user selects the vehicle type from one of the three Finally, exploring decentralized insurance on the blockchain
options as of yet (mini, sedan or SUV). holds promise for more accessible coverage. These
4. The driver with the relevant matching vehicle receives advancements position our dApp as a comprehensive and user-
notification on their dashboard for a ride and they can centric platform at the forefront of revolutionizing the ride-
accept it. This then triggers the ride block. hailing landscape.
5. The driver reaches the pickup location and picks up the
rider. Both mark this event in their app.
6. At the destination, both mark that they have reached the
destination in their app as well. This triggers the final VIII. CONCLUSION
event for the smart contract and the ether is transferred With our research on various prevalent ride hailing services, we
from the rider’s wallet to the driver’s wallet. have realized that these services have a lot of potential for data
7. After the ride is completed, the transaction is stored loss or data theft due to their centralized way of storing private
securely in the blockchain ledger, specifically on the data. Also, the involvement of the companies as a middleman is
Ethereum platform in our project. very high. In order to reduce this overinvolvement of the
middleman, we have made this system. This project is a
complete overhaul of the current ride hailing services. During
VI. RESULTS the process of developing this application, we chanced upon
Here we present some screenshots from our app. Fig 2 shows various existing works in this domain. We learnt a lot from
our user registration page while Fig 3. shows how a user can these works and came up with our own system. We deliver a
select their pickup and destination which generates appropriate completely decentralized ride hailing system which eliminates
polylines to provide visual feedback. Similarly, pages for the need of a middleman in the ridesharing industry. This app
confirming the ride, making the payment and rating the users allows riders to choose a pickup and drop-off location
are also present. throughthe web app user interface and allows them to book
rides. We store this ride data on the blockchain. We also use
MongoDB as a secondary storage medium for data whichis
trivial for storage on the blockchain. This is a feature of our
application which makes us stand out from the existing
technologies.

REFERENCES
[1] Blockchain to Disrupt Uber: Entering Ridesharing Industry,
LeewayHertz. Retrieved on 1st June, 2022 from
https://www.leewayhertz.com/blockchain-disrupting-uber-platform/
[2] RideX | Decentralized taxi experience on Ethereum Blockchain, UX
Fig 2: Registration Page Planet. Retrieved 1st June, 2022 from https://uxplanet.org/ridex-taxi-
service-on-ethereum-blockchain-fecee1879a23
[3] PeerPool: A Decentralized Ride Sharing Application. Retrieved 1st
June, 2022 from https://github.com/adityakeerthi/PeerPool
[4] DeRiS: Blockchain based Ride Sharing. Retrieved 1 st June, 2022 from
https://github.com/Nirvan66/DeRiS
[5] Kudva, Sowmya, Renat Norderhaug, Shahriar Badsha, Shamik
Sengupta, and A. S. M. Kayes. "Pebers: Practical ethereum blockchain
based efficient ride hailing service." In 2020 ieee international
conference on informatics, iot, and enabling technologies (iciot), pp.
422-428. IEEE, 2020.
[6] Baza, Mohamed, Noureddine Lasla, Mohamed MEA Mahmoud,
Gautam Srivastava, and Mohamed Abdallah. "B-ride: Ride sharing
with privacy-preservation, trust and fair payment atop public
blockchain." IEEE Transactions on Network Science and Engineering
Fig 3: Journey Route Page 8, no. 2 (2019): 1214-1229.
[7] Omar Dib, Kei-Léo Brousmiche, Antoine Durand, Eric Thea, Elyes
Ben Hamida. Consortium blockchains: Overview, applications and
VII. FUTURE SCOPE challenges. International Journal on Advances in Telecommunications,
IARIA, 2018.
We plan to move beyond a basic rating system towards a [8] V. Buterin et al., “Ethereum white paper: a next generation smartcontract
nuanced reputation system leveraging blockchain for tamper- & decentralized application platform,” First version, 2014.
proof data storage. Integration of in-app payments via [9] K. Christidis and M. Devetsikiotis, “Blockchains and smart contracts
established gateways or cryptocurrency options, combined forthe internet of things,” Ieee Access, vol. 4, pp. 2292–2303, 2016.
with machine learning for optimized driver allocation, will [10] Benet, J. (2014). IPFS - Content Addressed, Versioned, P2P File
enhance user experience. Furthermore, expanding beyond cars System. ArXiv, abs/1407.3561.
to include bikes, scooters, and public transportation caters to a
wider audience and promotes sustainability.