Cyber Security and Applications 3 (2025) 100069

Contents lists available at ScienceDirect

Cyber Security and Applications

journal homepage: http://www.keaipublishing.com/en/journals/cyber-security-and-applications/

An effective steganographic technique for hiding the image data using the
LSB technique
Rasmita Panigrahi, Neelamadhab Padhy∗
School of Engineering and Technology, Dept. of Computer Science and Engineering, GIET University, Gunupur, Gobriguda, Odisha, India

a r t i c l e i n f o a b s t r a c t

Keywords: Steganography is the art and science of writing secret messages so that neither the sender nor the intended
Steganographic techniques recipient knows there is a hidden message. Data hiding is the art of hiding data for various reasons, such as keeping
Encode, Decode private data, secure, confidential data, etc. With increasing data exchange over a computer network, information
Least significant bits
security has become a significant issue. There are many methods used for data hiding, and steganography is a
Stego image watermarking
well-known technique. Steganography is the art of invisible contact and science. Steganography is the process
QR Code
Embedding through which the presence of a message can be kept secret. The objective of this paper is to hide data using
the LSB (Least Significant Bit) technique into images that can be detected only by the specified user. We have
developed a user-friendly GUI such that it can be used with the utmost ease. This paper is motivated to hide the
message stated by the user in the dialog box given within the picture. The secret text is converted to the ciphertext
to make it more stable. The sender selects the cover image, and it is used to generate the secured Stegno image,
which is identical to the cover image. With the support of a private or public communication network, on the
other hand, the stegno image can be saved and sent to the designated user, i.e., the recipient downloads the
stegno image and can retrieve the secret text concealed in the stegno image using that same application. As for
the watermarking, we have visible and invisible we have used the same LSB technique. In visible watermarking,
text or image is embedded in the cover image, which can be noticed easily. As for invisible watermarking, some
specific text is inserted into an image, and while retrieving it, it generates a QR code, which can be scanned to
get the watermarked text. We used the three different types of cover images i.e. Gray, and RGB also estimated
the performance metrics. SNR, MSE, and PSNR the three performance metrics are used, and found that PSNR
achieved good results i.e.,71.4733. The RGB image with the hidden text is achieved up to 77.6697

1. Introduction the listener cannot hear the masked quieter tone. The primary goal of
steganography is to facilitate secret communication in an organization
The growth of electronics has led to the digitization of all documen- or between two users. An anonymous person cannot access the informa-
tation, videos, and audio. This scenario has increased the need for the tion by having a look at the cover file. Steganography is different from
security and reliability of documents, audio, or videos to maintain pri- cryptography. It is employed to hide the data inside another cover ob-
vacy, avoid piracy, and mass production reproduction. This varies from ject, whereas, in cryptography, data is encrypted. If the information is
organization to organization or individual to individual. Cryptography extracted in cryptography, then the exact data would again be difficult
and steganography are used to achieve the above requirements. Today’s to obtain.
digital media provides reliable and easy ways to edit data, and these data Cryptology is the way to send the encrypted message further needs
are needed to be delivered safely over a secure network. Steganography to be decrypted in the form of the original. Since the attacker would not
is a process of hiding data in the bits of cover objects like a graphic or an recognize the hidden message during a communication, this mechanism
audio file. Steganography provides a safe way to communicate privately ensures confidentiality.
because the presence of information in the cover item is challenging to However, the unreadable type can alert adversaries to the fact that
detect. Media files are suitable for steganography due to their large size. a hidden communication is taking place, making the data vulnerable to
Data can be inserted into a file using human perception. The audio file active attacks such as data blocking or alteration. As a result, steganog-
uses frequency masking on the tones with the same frequencies, and raphy mechanisms have been introduced as a way to get around the lim-

Peer review under responsibility of KeAi Communications Co., Ltd.

∗
Corresponding author.
E-mail address: [email protected] (N. Padhy).

https://doi.org/10.1016/j.csa.2024.100069
Received 6 May 2024; Received in revised form 13 July 2024; Accepted 2 August 2024
Available online 14 August 2024
2772-9184/© 2024 The Authors. Publishing Services by Elsevier B.V. on behalf of KeAi Communications Co., Ltd. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
R. Panigrahi and N. Padhy Cyber Security and Applications 3 (2025) 100069

itations of encryption techniques. Steganography is the process of em- issues. Swain, G. et al. [11] proposed an algorithm that allows han-
bedding hidden data into a cover medium to create a stego medium. The dling high-capacity stegno images which is the LSB technique. Their
stego medium must be made to look exactly like the cover medium.et approach is to divide the image into non-overlapped 3 × 3-pixel blocks
al., [13] used the DCT technique in steganography. They exhibited the after they have used the LSB technique for every block subsequently.
correlation and concluded, that if the correlation is high then proved the Nolkha, A. [12] has done a systematic literature survey. They mainly
stronger compaction. They discussed more about the data hidden con- focus on how LSB is used in the stegoimage. They have taken a color
cept. They also discussed how segmentation helpful for steganography image because it holds a large amount of secret data. Speech steganog-
Major Contribution: raphy techniques are divided into two categories based on the hiding
domain: time-domain schemes and frequency-domain schemes. Secret
1. We have developed an LSB-based steganographic technique messages are encoded in time-domain speech samples in a time-domain
2. Created one GUI-based application and Outlined the process of hid- scheme. On the other hand, frequency-domain systems embed the hid-
ing a user-specified message within an image, converting it into ci- den message in a specific part of the speech signal’s frequency spectrum.
pher text for enhanced security, and generating a stegno image that Several speech concealment systems have recently been published in the
appears identical to the original cover image. literature [14-16].
3. Described properly the process of securely transmitting the stegno
image over private or public communication networks to the desig-
3. Steganography and types
nated recipient, who can then retrieve the hidden message using the
same application.
Steganography attempts to hide information in the covering data in
4. Used the performance metrics, with results demonstrating high im-
such a way that non-participating persons can not recognize the pres-
age quality retention.
ence of this information by analyzing the detection of data. In contrast
This article contains 8 sections. Section 1 provides a details back- to watermarking, steganography is not intended to prevent opponents
ground of steganography and their insights. Sections 2 and 3 dis- from removing or changing the hidden message found in the cover data
cuss the state-of-the-art of steganography and its types. Sections 4 and from the secret material. Still, it emphasizes that it is steganography is
5 give the outline of the digital watermarking system and its procedure. especially essential for applications where confidential information can
Section 6 discusses how the system is designed for the application and not be encrypted to secure communication.
Section 7 explained briefly the algorithms required to be developed for
the application. Sections 8 and 9 discussed the system design and per- 3.1. Text steganography
formance metrics evaluation
It is a steganography process where information is concealed in text
2. Literature review files. The data is hidden by manipulating the spaces and tabs after nth
letters or nth lines. Text steganography is attained by changing the text’s
Alma, Wardhani, et al. [1], have developed the program used to send format or modifying the characteristics of the characters. This approach
the original bar code and generate the stegoimage barcode using the aims to create change, which is reader-friendly and unnoticeable. These
least significant steganography technique. Such two images were com- characteristics are a bit conflicting with each other. To achieve steganog-
pared and checked with performance metrics like MSE, RMSE, eans, raphy, various methods can be used for steganography, and these are
and PNR. Ramesh, M. et al. [2]. has developed the hybrid steganogra- • Line-Shift Coding: A text file can be modified by vertically shifting
phy method, which allows reading the QR code and converting it into the positions of the lines to encode the file uniquely.
the discrete wavelet transformation form. In this article, they have used • Word-Shift Coding: A text file can be modified by horizontally shift-
two different techniques; these are frequency domain encoding and de- ing the position of the words in each line to encode the file uniquely.
coding. Their objective was to hide the information. Chitradevi, B. et al.
[3]. discussed the data hiding techniques by using LSB steganography.
This paper presents a short idea for image steganography, which uses the 3.2. Image steganography
LSB algorithm to hide the data in an image. Zhu, J. et al. [4] have used
the deep learning technique for data hiding. They have discussed the Steganography technique in which information is hidden inside an
adversarial, which helps to improve the encoded image video appear- image. The image used for protecting the data is called a cover image.
ance. The novel Convolutional Network is used to learn to encode a rich In this technique, the intensities of the pixels are modified according to
by using invisible disturbance Relevant quantities of details. Bal, S.et al. the secret data. In digital steganography, images are preferred to hide
[5]. discussed the image watermarking technique by using LSB and bit data because it consist of many bits that digitally represent the image.
pair similarity. They have the symmetric key cryptography technique
for data security purposes. Jung, K. et al. [6]. discussed the method of 3.3. Ssteganography in image
hiding the data using the LSB technique also discussed a semi-reversible
data-hiding method. The LSB technique is used to insert secret data. Ba- In image steganography, the secret message or data is hidden inside
matraf, A. et al. [7]. suggested a new kind of algorithm called embed- the images by altering their visible properties. For example, the cover
ding and extraction. They have used peak signal-to-noise ratio (PSNR) image can be modified in the noisy with color variations to remain un-
for reading the quality images. Duan, X. et al. [8] have discussed about noticeable of the modifications. The Least Significant Bits(LSB) method
the PSNR and SSM. On their finding, they obtained values around 40 dB is the most common method to alter the images, and it also consists
and 0.96 of processes like masking, filtering, and transformation. These methods
They used both cryptography and deep neural network techniques are used according to image quality. Image steganography is performed
to enhance image steganography. Pichardo-Méndez et al. [9] studied for images in such a way that the embedded data can be retrieved eas-
the LSB mechanism to hide the information from the digital image. The ily. Image steganography refers to embedding data, i.e., text, images,
author YiZhang et al., [10]. They have used the proposed algorithm or audio or video files. Our project aims to use image steganography
where more statistical inference is used. They have discussed how the to use spatial domain techniques to achieve text hiding and visible and
stego image suffers during JPEG compression. Due to this reason, they invisible watermarking. Only a particular decoding technique will re-
have developed a methodology called “compression-resistant Domain trieve the secret message. Python incorporates both the encoding and
Constructing + RS-STC Codes, “ which resolves the JPEG compression decoding techniques.

2
R. Panigrahi and N. Padhy Cyber Security and Applications 3 (2025) 100069

E is an encoder feature, it takes an image I and a signature S, and it

produces a new model called a watermarked image I’, that is to say.
E(I, S) = I′ .

5.2. Decoding

A decoder function D takes an image J (J may be a watermarked or

unwatermarked image, and probably corrupted) whose ownership is to
Fig. 1. Watermark encoding procedure. be determined and recovers from the image a signature S’. An additional
picture can also be used in this process that also the original and unwa-
4. Digital watermarking termarked version of J. This is because specific coding systems can use
the authentic images in the watermarking process to ensure extra ro-
Digital watermarking or watermarking is a way and method to hide bustness against deliberate and unintentional pixel corruption.
information, and they can be text or media files. Watermarking is a pro- Mathematically.
cess in which a message is hidden in a digital object like text, pictures
D(J, I) = S′
or videos, or images and can be retrieved Steganography is another type
of watermarking where information in the digital object is shielded ex- The below mentioned Fig. 2 describes the watermark decoding pro-
istence is not visible. The currency of India is an example of watermark- cess
ing. An image goes through an embedded processor with a watermark The Watermarking method used can involve unique approaches de-
in the general watermarking process, producing a watermarked image. pending upon the way the watermark is embedded and relies on the
behavior of the watermarking algorithm. In some cases, the actual wa-
5. General procedure for watermarking termark is extracted in the same form, called watermark extraction. In
other schemes, only the presence of the specific watermarking signal is
Watermarking is a process in which data is called a watermark or detected, called watermarking detection.
tag or digital signature or label into a digital object so that watermark is
discovered and fetched later to make an attestation about the object. The 6. System design
object may be a video or image or audio. Generally, the watermarking
algorithm consists of three parts: The below-mentioned Fig. 3 represents the schematic diagram of
visible watermarking techniques. A visible watermark is a transparent
• Watermark
overlay in a picture and can be seen, by the viewer. Visible watermark-
• Encoding algorithm
ing is used to display ownership and to protect copyrights. Whereas an
• Decoding algorithm
invisible watermark is inserted in the data so that the changes made to
Watermarks are typically used for copyright protection of objects by the pixel values are not seen perceptually. Invisible watermarks are used
their owners. Every owner has a unique watermark or and can use dif- as evidence of ownership and for detecting misappropriated
ferent watermarks for different objects. The encoding algorithm embeds The above-mentioned Fig. 4 is the complete layout of our applica-
the watermark on the object, and the decoder algorithm authenticates tion. The working principle of the application is as below:
the object by determining the integrity and owner (Fig. 1). The user will supply the source file from the editor, then the python
interpreter will compile it into the byte conde through the VM connect-
5.1. Encoding ing to the different libraries and finally, it gives the output.

The below-mentioned Fig. 2 shows that the original image will be 7. BRIEF algorithm implementation
converted into the watermarked image with the help of the embedding
procedure. The Figure below describes the watermark encoding process: 7.1. Least significant bit (LSB)
Let us denote an image by I, a signature by S = {s1, s2, …sn} labeled
the watermarked image. The least significant bit is the most commonly used technique in
image steganography. LSB technique involves altering pixel intensities
by a small value, which results in a negligible change in the image,
which remains unnoticeable by the human eye.
There are two different methods to achieve image steganography:

• Spatial method
• Transform method

But for our project, we are using the Spatial method.

7.2. Spatial method

In the LSB method, the spatial method is most commonly used. It is

a common, simple process for hiding or embedding information on an
object or cover file. In image steganography, the LSB method is used
for each pixel. An image has three components (Red, Green, and Blue).
The information of the pixel is stored in an encoded format in one byte.
To hide data in an image, the first criterion is that the text size should
be less than equal to the size of the image used as a cover image. The
resulting image in this process is vulnerable to cropping and noise. In
this method, the most significant bits of the text are stored in the least
Fig. 2. Watermark decoding process. significant image bits. Pixels in an image are stored as bits in the image.

3
R. Panigrahi and N. Padhy Cyber Security and Applications 3 (2025) 100069

Fig. 3. Schematic Design for visible watermark.

Each pixel is 8 bits in the gray image, and for a color image, each pixel 7.4. Decoding Procedure
is 24 bits. Our project is only applicable to colored images. The Human
Visual System cannot detect modifications of pixel intensities when the Table 2
least significant bit is modified. LSB is used as an advantage to hide Pseudocode For Encoding Procedure.
information in the image with minimum differences in the picture. 1. Extracting all the pixels from the stego image.
2. Taking 3 pixels at a time and storing the LSB of each color in an array.
3. Checking third-pixel LSBs. If it is 0 means, it is the end of the text message else
7.3. Encoding Procedure continue to store the LSB in the array.
4. After storing all the LSB values, it is transformed into a ciphertext character.
5. Then the ciphertext is converted back to the original text using the same
Table 1 previous key.
Pseudocode for encoding procedure.

1. Extracting all the pixels from the given image.

2. Extract all the characters from the given text.
3. Next phase, converting the given text into cipher text for more security using a During implementation, we have used the databases(MySQL, SQLite,
specific key. Firebird, Oracle) and the test frameworks (Robot, PyTest, Unittest,
4. Take 3 pixels and one character from the ciphertext. Each character is of 8 bits, DocTest, Nose2, Testify), web scrapingMechanize, BeautifulSoup), Im-
so replacing the LSB of each color, i.e., RGB, with every bit of the character. age processing (Scikit-image, OpenCV, SciPy, Pillow). For Data Analyt-
5. If all characters are present, the LSB of the third color of the third pixel is 0, or
ics, we have used toolkits like -NLTK, SciPy, and NumPy, apart from the
1 if all the characters are left.
6. Each character is stored in three pixels, and the LSB of the third pixel Machine learning algorithms used. The above-mentioned Tables 1 and
determines the end of the message. 2 for the pseudocode for encoding and decoding procedure.
7. As the text embedding finishes, a stego image is generated, which is identical to The simple techniques hide the bits of the text directly into the LSB
the cover image.
of the image or cover image in a linear sequence. Modulation of the LSB

Fig. 4. System Design.

4
R. Panigrahi and N. Padhy Cyber Security and Applications 3 (2025) 100069

will not result in noticeable human differences because the change is Cryptography) technique, the watermark is encrypted, which protects
very small. In this method, the bits of the pixels are used. Hence it is against threats. The two different algorithms proposed for image en-
necessary to use a lossless compression format like PNG, or the embed- cryption and decryption (Table 3).
ded information will be lost in the transformation. At first, we have to add the plain text which is to be encrypted then
Let 3 pixels are given below in the form 2D matrix; each pixel is of we have to use the encryption algorithm (AES). So, that it converts the
24 bits. plain text to the ciphertext, now we have to decrypt the message so
again we have to use the decryption algorithm so that it can convert the
(01,100,111 11,100,001 11,001,100) ciphertext into the plain text again and then send it to the receiver.In
(00,101,111 11,000,000 10,101,001) this context,confidentiality and data integrity are the important and its
(11,010,000 00,100,001 11,111,001) like as a safefuard against the unauthorised access and use.The two es-
sential things for data sequrities are cryptography and steganography.
Character ‘A’, binary value equals 10,000,001, is inserted into the
The cryptography providesconceals the message existence and other
2D matrix of the 3 pixels, the following grid results:
technique is allowed to modity the contents . The data is encrypted
and sent after being converted into some other form of gibberish us-
(01,100,111 11,100,000 11,001,100)
ing cryptography .In steganography, a picture file is delivered with data
(00,101,110 11,000,000 10,101,000)
contained inside. This study emphasises how combining steganography
(11,010,000 00,100,001 11,111,000)
and encryption techniques can improve the security of communication
All the bits of ‘A’ are stored in the LSB of each color, but LSB of the The above-mentioned (Figs. 7-9) means encryption and decryption tech-
third color of the third pixel is made 0, it acts as a delimiter, which niques which are implemented and demonstrated. Using multiple layers
signifies the end of the message data. Only four bits are changed out of of LSB embedding, such as multi-bit LSB, increases the security of the
72 bits to insert the character successfully in the above case. On average, hidden data. However, it also makes the steganography more detectable
less than half of the bits are required alteration to embed the message and can reduce the quality of the host image. Our analysis shows a bal-
in the cover image. The modification made to the LSB is too small to be ance between security and image quality must be maintained (Figs. 5–
noticed by the human eye, so the message is hidden efficiently. 9).
For Example, Text=” Steganography” is embedded in the below
cover image:

Fig. 6. Stego Image.

Table 3
Fig. 5. Cover Image.
Proposed algorithm for watermarking.

Several mechanisms exist for disseminating information in various Proposed algorithm for watermarking
media. LSB coding is used to manipulate images or compression tech- The required input: We will provide the grayscale image
nologies that affect image relations. Basic replacement systems attempt 1: Take the gray picture W.
to encode hidden information by substituting inconsequential portions 2: For every pixel (Pi) of W, produce the decimal value.
of the cover with secret message bits. If the receiver understands where 3: Determine the accurate binary representation for each Pi..
4: Invert the binary number Bv of 8 digits to get Rv.
the sensitive information is embedded, he can extract it. A passive at-
5: Find the Key (Ke) of a four-digit divisor.
tacker is unaware of the modest alterations made throughout the em- 6. Next divide the reversed Rv number with the Ke divider.
bedding process. 7: The rest and quotient are stored next in an 8-bit array. If necessary, add zeros to
Incorporating the encryption algorithm in further, it enhances the the left of the remaining bits and quotient bits to complete an 8-bit string.
Output: The proposed algorithm produces encrypted data (ED).
security of the proposed mechanism.By using the SKC (Symmetric Key

5
R. Panigrahi and N. Padhy Cyber Security and Applications 3 (2025) 100069

Fig. 7. Steganography Encrypt Image selection.

Fig. 8. Message selection.

Fig. 9. Decrypted message.

The above-mentioned Tables 4 and 5 are represented for proposed image generates the ciphertext with the help of a secret text message.
watermarking and decrypted watermarking algorithms Then the ciphertext used the LSB technique to convert into a stegno
image. On the receiver side, the user needs to select the stegno image,
8. System design then after the novel, the LSB decryption technique is used to transform
it into the ciphertext to get the secret text message. The above Fig. 10
8.1. Use case diagram is used for text hiding using a use case diagram.

A Use Case Diagram is a representation of user interaction with the 8.2. Invisible watermarking
system. When retrieving, the user selects the stego image after initially
selecting the cover image, which is transformed into a stegno image. The user selects the cover image at the beginning, then from the
Then, decryption is applied to the segno image. From the cover image cover image generates the text data to produce the QR code. The QR
to the stegno image, it has undergone several steps. First of all, the cover code is used for invisible watermark implementation purposes. Finally,

6
R. Panigrahi and N. Padhy Cyber Security and Applications 3 (2025) 100069

Fig. 10. Use case diagram for Text Hiding.

Table 4 veloped a model that allows the watermark implementation module to

Proposed algorithm for decryption watermarking. extract the image for sharing purposes.
Proposed algorithm for message extraction
8.4. The embedding algorithm
The objective of this algorithm is to input the stegno image as well as the secret key
(stegno-key), and our implementation part shows the hidden message. Thus, this
proposed algorithm is called a message extraction algorithm. The embedding uses a series of pseudo-random data or grayscale im-
1. Open the image file Stego in reading mode, and read the RGB from the image file ages as secret data, together with a grayscale cover picture. A grayscale
Every pixel color.
stego image is the end outcome. Let the remainder and quotient arrays
2. Extract the host images which contain the Red component.
3. Read every pixel’s last bit.
be A and B, respectively. The size of each array may not exceed half that
4. Initialize the random key, which gives the message bits location in the red Pixels, of the cover image.
which are randomly inserted. The above algorithm mentioned in Table 6 is the proposed embed-
5. Select the pixels for decoding and Extract the red pixels LSB value. ding algorithm that is used in our application
Output: At the end of this algorithm, our system will provide the secret
message behind the stegno image
9. Iimplementation setup
Table 5
We employ the feature similarity (FSIM) and peak signal-to-noise
Proposed algorithm for message extraction.
ratio (PSNR) metrics to assess the effects of image segmentation. PSNR
Proposed Algorithm for Decryption Watermark is used to compute the PSNR of the segmented image and the original
Input: Take the two parameters: image.
The Encrypted image data (ED) along with the encryption key (Ke) The PSNR index is determined using the following formula:
1: To generate F, multiply the key (Ke) by the quotient bits of encrypted data (ED). ( )
255
2: To get G, add the remaining bits of the encrypted data (ED) with the result 𝑃 𝑆𝑁𝑅 = 20𝐿𝑂𝐺 (𝑑𝐵 ) (1)
produced in step (F) above. 𝑅𝑀𝑆𝐸
3: When the result generated (G) is not an 8-bit number in the previous step, i.e., Where
step 2, we need it to make it an 8-bit number. √
∑𝑁 ∑𝑛
(𝐼(𝑖, 𝑗 ) − 𝐼̂(𝑖, 𝑗 ))
4: In order to obtain the decrypted data (DD), flip the G number. 2
Output After executing, we will get the decrypted image 𝑖=1 𝑗=1
𝑅𝑀 𝑆𝐸 = (2)
𝑀 𝑋𝑁
The parameters M and N are termed as the image size, and I is the
it converts into a hidden stegno image. The other end receives the stegno color of the original image
image in the form of watermarked; then, the extraction algorithm gen- The Feature Similarity (FSIM) method is used to determine how
erates the QR code. Finally, scan the generated QR code. The entire pro- structurally identical the original and segmented images are. FSIM is
cess is called an invisible watermarking technique. The above Fig. 11 is described as follows:
∑
used for the Invisible Watermarking use case diagram. 𝑥∈𝛀 𝑆𝐿 (𝑥).𝑃 𝐶𝑚 (𝑥)
𝐹 𝑆𝐼𝑀 = ∑ (3)
𝑥∈𝛀 𝑃 𝐶𝑚 (𝑥)

8.3. Visible watermarking The above equation Ω indicates the entire color image and the FSIM
can be represented as the RGB color.
The below-mentioned Fig. 12 is used for Visible Watermarking. In 1 ∑ ( )
𝐹 𝑆𝐼𝑀 = 𝐹 𝑆𝐼 𝑀 𝑥0 , 𝑦0 (4)
this case, the user submits the data as a text or image. We have de- 𝑂 𝑜

7
R. Panigrahi and N. Padhy Cyber Security and Applications 3 (2025) 100069

Fig. 11. Use case diagram for Invisible Watermarking.

Table 6
Proposed embedding algorithm.

Proposed embedding algorithm

1. The maximum and minimum values in the secret data are determined using the
decimal number between 0 and 255.
2. Subtract the greatest value from the minimum and each value of secret data.
3. After dividing the outcome by 32, store the quotient in array A, and divide the
remaining amount by 8.
4. In array B, store the quotient from the second division and the remainder in array
C.
5. Divide the greatest value by 32 and 8 and save the values in the first pixel of A, B,
and C, correspondingly.
6. By inverting the values of these five LSBs, you can embed quotients of the
maximum value in the first five LSBs of the cover image’s first half.
a) If B(1, 1) = 1, then flip the value of the first LSB.
b) Reverse the second LSB if B(1, 1) equals 2.
c) Reverse the first and second LSBs if B(1, 1) equals 3.
d) Reverse the third LSB if A(1, 1) equals 1.
e) Reverse the fourth LSB if A(1, 1) equals 2.
f) Reverse the third and fourth LSBs if A(1, 1) equals 3. Invert the value of the
fifth LSB if A(1, 1) equals 4.
g) Invert the values of the third and fifth LSBs if A(1, 1) equals 5.
h) Invert the values of the fourth and fifth LSBs if A(1, 1) equals 6.
i) Invert the value of the third, fourth, and fifth LSBs if A(1, 1) equals 7.
Fig. 12. Use case diagram for visible watermarking. 7. As mentioned in the previous step, embed the remainder of the highest value in
three LSBs of the first pixel in the second half of the cover image.
8. Use the fourth bit of this pixel’s value as an indicator for R. If this value is less
than or equal to 127, the four LSBs of each pixel are inverted in the first step of the
The above Eqs. (2-4) are used for feature similarity purposes. The embedding process, otherwise, five LSBs are inverted. Because dividing R values in
the range of 0 to 127 by 32 requires two bits for representation while dividing R
term X0, y0 reflect the original image’s oth channel and the segmented
values above 127 requires three bits.
image’s oth channel, where o is the channel amount. Data hiding has 9. Apply the best LSBs algorithm to the pixels you got in steps 5 and 7.
an effect on the video bit rate in addition to its effect on the percep- 10. To embed each pixel in arrays A, B, and C in pixels of the cover picture, repeat
tual quality of the video. In general, as compared to the original com- steps 5 and 6.
11. For each pixel in the first half, invert certain inverted bits once more. One or two
pressed video, the bit rate of the marked compressed video would in-
inverted bits will be inverted again in one of two circumstances. The value of R will
crease. To assess performance further, the bit-rate increase ratio (BIR) is determine this.
introduced, which represents the bit rate created by the proposed data 12. After using the optimal LSBs approach, each pixel now has two values; choose
embedding encoder relative to the bit rate generated by the original en- the one that is closest to the original value and return 0 or 1 as an indicator of which
coder. Because zero QTCs are not affected during data embedding, the invert was chosen.
13. According to this indicator, invert the fourth bit of the pixel in the second
bit-rate increment remains within an acceptable range.
portion.
14. Apply the best LSB technique to the pixel from the previous step.
𝐵𝑅𝑒𝑚 − 𝐵𝑅𝑜𝑟 15. The stego picture is formed after embedding everything in the cover image, and
𝐵 𝐼𝑅 = × 100% (5)
𝐵𝑅𝑜𝑟 it is then sent to the receiver.

⎧ ( ) ( ) ⎧ ( ) ( )
⎪ 𝑥𝑖′ 𝑖𝑓 𝑚𝑖 = 𝐿𝑆𝐵 𝑥𝑖 , 𝑚𝑖+1 = 𝑓 𝑥𝑖, 𝑥𝑖+1 , ⎪ 𝑥𝑖+1, 𝑖𝑓 𝑚𝑖 = 𝐿𝑆𝐵 𝑥(𝑖 ,) 𝑚𝑖+1 = 𝑓 𝑥(𝑖, 𝑥𝑖+1 ,)
( ) ( )
⎪ 𝑥𝑖′ 𝑖𝑓 𝑚𝑖 = 𝐿𝑆𝐵 𝑥𝑖 , 𝑚𝑖+1 ≠ 𝑓 𝑥𝑖, 𝑥𝑖+1 , ⎪𝑥 1, 𝑖𝑓 𝑚𝑖 = 𝐿𝑆𝐵 𝑥𝑖 , 𝑚𝑖+1 ≠ 𝑓 𝑥𝑖, 𝑥𝑖+1 ,
𝑦𝑖 = ⎨ ( ) ( ) (6) 𝑦𝑖 = ⎨ 𝑖+1 ± ( ) ( ) (7)
⎪𝑥𝑖, − 1, 𝑖𝑓 𝑚𝑖 ≠ 𝐿𝑆𝐵( 𝑥𝑖) , 𝑚𝑖+1 = 𝑓 𝑥𝑖 − 1, 𝑥𝑖+1 ,
( ) ⎪𝑥𝑖+1′ 𝑖𝑓 𝑚𝑖 ≠ 𝐿𝑆𝐵 (𝑥𝑖 ), 𝑚𝑖+1 = 𝑓 (𝑥𝑖 − 1, 𝑥𝑖+1 ),
⎪ 𝑥𝑖, + 1, 𝑖𝑓 𝑚𝑖 ≠ 𝐿𝑆𝐵 𝑥𝑖 , 𝑚𝑖+1 ≠ 𝑓 𝑥𝑖 − 1, 𝑥𝑖+1 , ⎪𝑥𝑖+1′ 𝑖𝑓 𝑚𝑖 ≠ 𝐿𝑆𝐵 𝑥𝑖 , 𝑚𝑖+1 ≠ 𝑓 𝑥𝑖 − 1, 𝑥𝑖+1 ,
⎩ ⎩

8
R. Panigrahi and N. Padhy Cyber Security and Applications 3 (2025) 100069

Table 7
Characters of the two approaches are compared.

Techinques Imperceptibility Robustness Capacity Tamper Resistance

∗
Simple LSB H L H L
Pseudo-Random Encoding HH∗ L H H∗ ∗

The following terminology is used in the above-mentioned table and represented below as
follows:
H: High.
L: Low.
HH: Higher.
The single (∗ ) represents depending on the cover image utilized and the double(∗ ∗ ) used
the key as well as the random seed. During our test, we used a grayscale/RGB image as the
cover image.

Table 8
Pseudo-random encodings PSNR.

Cover Image Secret Message Stego-Image SNR(dB) MSE PSNR(dB)

Gray image Text message Gray image 69.5043 0.0563 71.4733

RBG image Text message sisbr 71.3649 0.031 77.6697
RBG image Image Hydrang 63.9812 0.0912 68.5311

Table 9
Least significant bits encoding PSNR.

Aspect SNR (dB) MSE PSNR (dB)

Gray image, Text message 59.5022 0.0673 71.4833

RGB image, Text message 71.4659 0.0312 77.7687
RGB image, Image 63.9921 0.0922 68.5321

Fig. 13. Comparision of pseudo-random encodings PSNR.

{ ( )
𝑥𝑖+1′ 𝑖𝑓 𝑚𝑖+1 = 𝑓 𝑥𝑖 .𝑥𝑖+1 ,
𝑌𝑖+1 = (8)
𝑥𝑖+1 ± 1, 𝑒𝑙𝑠𝑒
The above-mentioned table 9 is meant for performance metrics for
To get the message, we only need to get mi from the first pixel’s LSB the least significant bits encoding PSNR.It describes the summarizing the
and mi+1 from the embedding phase’s function. PSNR values and their performance metrics for our proposed model.
In LSB MR, each of the message’s two bits is hidden in the brightness The above Figs. 13 and 14 are used for comparison between the Least
of two continuous pixels with just one change. This approach can only Significant Bits Encoding PSNR and Pseudo-Random Encodings PSNR
be used to hide two bits. We expand LSB MR to hide three bits with only techniques. The Fig. 15 depicts the learning curves for MSE and PSNR.
one chance in this paper. As the number of modifications decreases, the The blue line indicates the MSE followed by the green line denotes as
PSNR metric rises, strengthening our method’s defenses against attacks the PSNR in dB.
to reveal the secret message. The LSB-based methods implement their The above mentioned Fig. 16 is SNR curve that shows that how SNR
changes at random, without taking into account the image’s statistics. values are different for combination of images as well as secret image
In [17], Huang and Ouyang suggested a method for selecting accept- (Figs. 17–22).
able embedding places in LSB. Smooth image regions, redundant image The second experiment will determine how well the suggested
regions, and fragile image regions were the names given to the picture steganography technique is at concealing information in speech signals
areas. A small alteration in the image’s weak sections makes the image from diverse languages. Finally, Google Translate is used to translate
identifiable, and the authors proposed that vulnerable regions be pro- the text to be spoken into various languages.
tected by avoiding embedding messages within them. In the above implementation, these are the following facts found.
The above two strategies will be implemented and evaluated Images can be altered by nosing, diluting, contrast changes, etc. There
(Tables 7 and 8). should be more than enough pixels of information to insert into the
We built the aforementioned two ways in MATLAB, and the above- image. There is a possibility that the message will be exposed while
mentioned picture steganography algorithms have their own set of someone is screaming; if more than two people have the steganogra-
strengths and weaknesses. As a result, deciding on the most appropri- phy app, the secret message may also be discovered. If someone sees
ate strategy to use is critical. As previously stated, numerous parameters the code or has access to the software, then they can manipulate the
can be used to assess the effectiveness of the steganographic system. The data to their discretion. Reverse engineering can be applied because
following are some parameters [18]: the application is written in Python, which is an open-source language.
Perceptibility is the degree to which embedding information dis- Only intended users must know about the working of this software to
torts the cover medium to an unacceptably high degree.The Capacity keep confidentiality. An intruder can suspect images by noise analysis
refers to the quantity of information that can be concealed within a and by some other parameter—protection from Data Alteration by the
given medium, considering the perceptibility of changes, over a speci- third party who is constantly searching for potential customers. In prac-
fied duration.stego medium in an attempt to delete, erase, or change the tical use, embedded data are weak in most steganography application
embedded data software. This brings up new possibilities for an information-alteration

9
R. Panigrahi and N. Padhy Cyber Security and Applications 3 (2025) 100069

Fig. 14. Comparision of least significant bits encoding PSNR.

Fig. 15. Learning curves for MSE and PSNR.

protection application, such as a Digital Certificate Document System. 9.1. Hiding the message and Extract the hidden message.
People can transfer their digital certificate data to any person in the
globe without risk of change, forgery, or tampering. Access Control Sys-
tem because we choose with whom we have to share our information.
Based on the current situation, digital content is increasingly circulat-
These are the following steps that we have done in our implemen-
ing, and we can cover our information for fair use by using the access
tation.
control mechanism by supplying an access key to extract information. E- STEP 1: In the above implementation, we conceal our secret
commerce for the life insurance and banking sectors and Media, which message and will make assured that it is properly encrypted so that
needs to be encrypted for data breaches and dealing with the customer’s someone would want to extract the message from the image, he
privacy, is a severe concern of the institution, so here steganography would not be able to decipher the secret message we have hidden.
comes in handy. Digital Watermarking, which may be visible or invis- In order to construct our secret key, we will first start nano by
ible to the human eye, provides authentication for the originality of typing nano file.txt, then pressing return, and then we will enter
the documents such as images and video. The reason behind using LSB our secret key before exiting the program.
steganography is to efficiently embed the data with minimal distortion STEP 2: When we send our message, we’ll use the symmetric
key algorithm (as implemented by gpg), which will allow both the
as a comparison to other techniques like DCT or DWT.LSB provides a
sender and the recipient to use a single key for both encryption and
good balance of simplicity as well as computational efficiency of our decryption.
application.

10
R. Panigrahi and N. Padhy Cyber Security and Applications 3 (2025) 100069

Figs. 16. SNR Curve.

Fig. 17. Main menu of Steganagraphy.

Fig. 18. Text in Image for convertion.

11
R. Panigrahi and N. Padhy Cyber Security and Applications 3 (2025) 100069

Fig. 19. Retrieve text.

Fig. 20. Invisible watermarking.

The similarity between the original cover speech signal and the stego
type gpg -c msg.txt to encrypt it. speech signal that contains the hidden data is referred to as the consis-
Step #3: It will prompt you for a password; simply type in the tency attribute. The Mean Absolute Error (MAE) and the Mean Square
one you like. Error (MSE) can be used to compare the similarity of two signals (MSE).
Step #4: Uploading this encrypted message to a hosting server In addition, a new metric, the Segmental Signal-to-Noise-Ratio, is used
and then copying the downloading link to include it in our image
in this paper to assess the accuracy of the stego signal (SNRSeg). The
is how we accomplish this task.
SNRSeg is a favored speech quality indicator that yielded more consis-
Step#5: To decode the hidden information contained within
the image tent results than traditional quality indicators like the MSE and MAE.
type decode –i “ IMG” -o “Target o/p format” The SNRSeg calculates the average of short segment Signal Noise Ratio
Step#6: We have successfully extracted the file from the im- (SNR) values. Calculating the difference between the two test signals
age; now, it is time to read the information contained within it. sample by sample yields the SNR.
type cat msg.txt to read the file.
Step#7: It is now necessary to decrypt the file that has been 𝑀−1 𝑁 𝑚∑+𝑁 −1
( ∑𝑁 2 )
10 ∑ 𝑖 𝑥 (𝑖)
downloaded. This program will prompt us to enter a password, 𝑆𝑁𝑅𝑆𝑒𝑔 = 𝑙𝑜𝑔10 ∑𝑁 =1 (9)
after which we will receive our secret message. 𝑀 𝑚=0 𝑖=𝑁𝑚 𝑖=1 (𝑥(𝑖) − 𝑦(𝑖))
2

The terms M and N are the number of segments and the length of
segments, respectively.The SNRSeg has produced more consistent re-
Identifying and extracting the hidden message Our recipient has suc- sults than the standard SNR metric, especially for waveform encoders.
cessfully received the image file, and it is now time to expose the con- This section is primarily used for the proposed methodology based on
cealed message contained within it. the extraction process. It can also be possible to extract the signals when

12
R. Panigrahi and N. Padhy Cyber Security and Applications 3 (2025) 100069

Fig. 21. Retrieve watermarking.

Fig. 22. Visible watermarking.

we provide the secret speech. We need to provide samples of speeches it should be noted that the approach is only useful for JPG and PNG
to retrieve similar kinds of embedding types. images as these require lossless compression techniques, as discussed
Steps above. It is also important to note that while steganography was once
The same splitting threshold value used at the embedder is used to undetected, with the various methods currently being used, it is easier
segment the obtained stego speech signal. to detect the presence and easier to obtain.
√
Each section is subjected to the DCT.
√
Since the magnitude contains the hidden data, only the magnitude Future scope
of the DCT is extracted.
√
To locate the hidden data, each segment is subjected to the 64-bit Steganography could allow for avoiding inaccuracies in the printing
quantization and scanning of media(data).: Can steganography be detected in image
files? That is a severe issue. A simple steganographic technique can be
10. Conclusion identified by merely analyzing the low-order bits of the bytes in the
image. However, if the steganographic algorithm is more complicated
It is observed that the results obtained in data hiding are very im- and the embedded data is scattered over the image, this is random or
pressive through the LSB steganographic substitution process because encrypts the information before embedding. Inline images are widely
it uses the simple fact that each image can be broken up into individ- used by the World Wide Web (www). Globally, there are millions of
ual bitplanes, each consisting of different information levels. However, pictures on different websites. An application can be created to act as a

13
R. Panigrahi and N. Padhy Cyber Security and Applications 3 (2025) 100069

web browser to retrieve images embedded in the web page. This stego- [4] J. Zhu, R. Kaplan, J. Johnson, L. Fei-Fei, Hidden: hiding data with deep networks,
web will work on top of the current Website and could be a way of in: Proceedings of the European conference on computer vision (ECCV), 2018,
pp. 657–672.
transmitting information secretly. The proposed embedded algorithm is [5] S.N. Bal, M.R. Nayak, S.K. Sarkar, On the implementation of a secured watermark-
designed for gray-scale images for secret data in the embedding. Our ing mechanism based on cryptography and bit pair matching, Journal of King Saud
application developed this proposed algorithm. In conclusion, the pro- University-computer and information sciences, 2018.
[6] K.H. Jung, K.Y. Yoo, A steganographic method based on interpolation and LSB sub-
posed LSB steganographic technique demonstrates high effectiveness in stitution of digital images, Multimed. Tools. Appl. 74 (6) (2015) 2143–2155.
hiding image data, as evidenced by the presented PSNR values. This [7] Bamatraf, A., Ibrahim, R., Salleh, M., & Mohd, N. (2011). A new digital watermark-
method can be effectively utilized in secure communication applica- ing algorithm using a combination of least significant bit (LSB) and inverse bit. arXiv
preprint arXiv:1111.6727.
tions, contributing to advancements in the field of cybersecurity and
[8] X. Duan, D. Guo, N. Liu, B. Li, M. Gou, C. Qin, A new high capacity image steganog-
digital data protection. raphy method combined with image elliptic curve cryptography and deep neural
network, IEEe Access. 8 (2020) 25777–25788.
[9] J.L. Pichardo-Méndez, L. Palacios-Luengas, R.F. Martínez-González,
Declaration of competing interest
O. Jiménez-Ramírez, R. Vázquez-Medina, LSB pseudorandom algorithm for
image steganography using skew tent map, Arab. J. Sci. Eng. (2019) 1–20.
The authors declare that they have no known competing financial [10] Y. Zhang, C. Qin, W. Zhang, F. Liu, X. Luo, On the fault-tolerant performance for a
interests or personal relationships that could have appeared to influence class of robust image steganography, Signal. Process. 146 (2018) 99–111.
[11] G. Swain, Very high capacity image steganography technique using quotient value
the work reported in this paper. differencing and LSB substitution, Arab. J. Sci. Eng. 44 (4) (2019) 2995–3004.
[12] A. Nolkha, S. Kumar, V.S. Dhaka, Image steganography using lsb substitution: a com-
CRediT authorship contribution statement parative analysis on different color models, in: smart systems and iot: Innovations
in computing, Springer, Singapore, 2020, pp. 711–718.
[13] M. Baziyad, T. Rabie, I. Kamel, Achieving stronger compaction for dct-based
Rasmita Panigrahi: Writing – original draft, Methodology, Concep- steganography: a region-growing approach, in: World conference on information
tualization. Neelamadhab Padhy: Writing – review & editing, Method- systems and technologies, Cham, Springer, 2020, pp. 251–261.
[14] R. Sridevi, A. Damodaram, S.V.L. Narasimham, Efficient method of audio steganog-
ology, Investigation, Formal analysis. raphy by modified lsb algorithm and strong encryption key with enhanced security,
J. Theoret. Appl. Informat. Techn. 5 (6) (2009).
References [15] J. Liu, K. Zhou, H. Tian, Least-significant-digit steganography in low bit-rate speech,
in: 2012 IEEE International Conference on Communications (ICC), IEEE, 2012,
pp. 1133–1137.
[1] R.W. Alma, Wardhani, S. Muhasyah, M Delina, The least significant bit of steganog- [16] M.A. Ahmed, M.L.M. Kiah, B.B. Zaidan, A.A. Zaidan, A novel embedding method to
raphy method for digital data protection in the barcode, AIP conference proceedings, increase capacity and robustness of low-bit encoding audio steganography technique
2169, AIP Publishing LLC, 2019. using noise gate software logic algorithm, J. Appl. Sci. 10 (1) (2010) 59–64.
[2] M. Ramesh, G. Prabakaran, R. Bhavani, QR-code image steganography, in: Proceed- [17] Q. Huang, W. Ouyang, Protect fragile regions in steganography LSB embedding, in:
ings of the second international conference on emerging research in computing, in- Proceedings of the 2010 3rd International symposium on knowledge acquisition and
formation, communication, and applications, 2014. modeling (KAM), Wuhan, China, 2010, pp. 175–178.
[3] B. Chitradevi, N. Thinaharan, M. Vasanthi, Data hiding using the least significant bit [18] A review of data hiding in digital images by E Lin, E Delp center for education and
steganography in digital images, Statistic. Approach. Multidisciplin. Res. 1 (2017) research information assurance and security purdue university, west lafayette, IN
144–150. 47907-2086.

14