www.nature.

com/scientificreports

OPEN A novel and efficient digital image

steganography technique using
least significant bit substitution
Shahid Rahman1, Jamal uddin2, Hameed Hussain1, Sabir Shah1, Abdu Salam3,
Farhan Amin4, Isabel de la Torre Díez5, Debora Libertad Ramírez Vargas6 &
Julio César Martínez Espinosa6
Steganography is used to hide sensitive types of data including images, audio, text, and videos in
an invisible way so that no one can detect it. Image-based steganography is a technique that uses
images as a cover media for hiding and transmitting sensitive information over the internet. However,
image-based steganography is a challenging task due to transparency, security, computational
efficiency, tamper protection, payload, etc. Recently, different image steganography methods have
been proposed but most of them have reliability issues. Therefore, to solve this issue, we propose an
efficient technique based on the Least Significant Bit (LSB). The LSB substitution method minimizes
the error rate in the embedding process and is used to achieve greater reliability. Our proposed
image-based steganography algorithm incorporates LSB substitution with Magic Matrix, Multi-Level
Encryption Algorithm (MLEA), Secret Key (SK), and transposition, flipping. We performed several
experiments and the results show that our proposed technique is efficient and achieves efficient
results. We tested a total of 165 different RGB images of various dimensions and sizes of hidden
information, using various Quality Assessment Metrics (QAMs); A name of few are; Normalized Cross
Correlation (NCC), Image Fidelity (IF), Peak Signal Noise Ratio (PSNR), Root Mean Square Error
(RMSE), Quality Index (QI), Correlation Coefficient (CC), Structural Similarity Index (SSIM), Mean
Square Error (MSE), Entropy, Contrast, and Homogeneity, Image Histogram (IH). We also conducted
a comparative analysis with some existing methods as well as security analysis which showed better
results. The achieved result demonstrates significant improvements over the current state-of-the-art
methods.

Keywords Image steganography, LSB, Image quality assessment metrics, Histogram analysis, Image,
Capacity, Robustness

During communication through different channels, information is more vulnerable to attackers, especially
when transmission takes place over the internet1. Secure information sharing over communication channels
is vital such as any leakage of private information may lead to persistent difficulties. The increasing need for
protecting information as being transmitted over communication channels has led to the discovery of various
mechanisms2. However, no single method can perfectly secure communication over the Internet, as each has its
advantages and disadvantages. Steganography is a popular method that involves the embedding of secret data
within cover objects like images, videos, audio, and text. The information hidden in such objects is not easily
detectable to the naked eye because steganography is a procedure that conceals the secret information within the
cover medium so that nobody can know about its existence3,4. In the abovementioned cover objects, images have
remarkable performance in concealing information that’s why it’s the most widely used option for steganography.
However, the basic research challenge in image steganography is achieving reliable criteria including capacity,
computation, robustness, temper protection, and transparency, as shown in Fig. 1. It is worthwhile to define
some key concepts relevant to image steganography. The payload; means a cipher text to be embedded within
the cover object. Security; means how much the original and stego image is strong against some statistical attacks

1Department of Computer Science, University of Buner, Swari 19290, KP, Pakistan. 2Riphah School of Computing
& Innovation, Riphah International University Lahore, Lahore, Pakistan. 3Department of Computer Science,
Abdul Wali Khan University, Mardan 23200, Pakistan. 4School of Computer Science and Engineering, Yeungnam
University, Gyeongsan 38541, Republic of Korea. 5Department of Signal Theory and Communications, University of
Valladolid, Valladolid, Spain. 6Universidad Europea del Atlántico. Isabel Torres 21, Santander 39011, Spain. email:
[email protected]; [email protected]

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 1

www.nature.com/scientificreports/

Fig. 1. Image steganography.

Fig. 2. Least significant bit (LSB).

such are cropping, tempering, scaling, etc. Perception; explains the nature of the cover picture after mounting
the mystery message. Temper protection; is a critical issue after securing the secret information within the cover
image because if someone tries to extract the message, temper protection; should be high enough to secure
the stego image against attackers. The computation shows the time taken by the extracting and embedding
process, and it should be minimized so the embedding and extracting process takes the least possible time5. The
evaluation of existing research works based on these criteria is important as it can lead us to identify the research
gap and suggest possible improvements. Most of the recent literature on image steganography is focused on one
criterion, but some have broken down the other and every stenographic method needs reliability between them6.
There is a tradeoff among different reliability criteria of steganography. Suppose we increase the limit of payload,
it will degrade the quality of the image and also affect the robustness7. The research challenge is two-fold: not to
decorate the quality of the cover object and to achieve reliable and best possible values for different parameters
of image steganography criteria8,9. Furthermore, appropriate object selection and optimum size of image pixels
and secret message size are vital.
These methods aim to hide a secret message within an image using a mysterious key. This ensures that the
collector cannot decipher the message without having the same key. The algorithm used LSB substitution and
value differencing concepts, resulting in a novel image steganography technique. The LSB technique is the most
widely used and popular technique for image steganography as shown in Fig. 2. For security, the algorithm uses
different concepts such as Magic Matrix and value differencing. Encrypting the message within the cover object
it reduces the error rate. The proposed method achieves reliability between the criteria up to acceptable limits6.
The outcomes show the significance and inspiration of the proposed calculation.

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 2

www.nature.com/scientificreports/

In LSB-based image steganography, the selected image is divided into corresponding channels shown in
Fig. 2. Each color intensity is then transformed into a corresponding binary representation. The secret message,
represented as ‘B’, is converted into ASCII values and further coded into dual values. The method uses the Red,
Green, Blue (RGB) color model because of its valuable properties. This model uses 24 depths per pixel, and each
color has 8 8-bit representations in the range of 0-25510. RGB is suitable for embedding more information because
having three colors and the combination of these colors can reproduce more possible colors11. Considering RGB,
we are working with a binary image having a range of 0-255 of eight bits’ representations where the left-most
bit is LSB and the rightmost bit is considered MSB. In addition, consider the range from 0 to 255, if we change
the MSB from 0, its impact is very high because it will change 99% intensity of the color. If we change LSB
from one to 1 or 0 to 1 then its impact will be very low say up to 2% change. Therefore, the proposed algorithm
used LSB concepts to embed the secret information, and according to the literature, up to 4 LSB bits, can be
embedded as secret information which is less susceptible to any naked human eye5,13. As mentioned earlier
state of art methods are not reliable because based on one or two parameters. So, this effort aims to develop a
novel reliable method that ensures reliability. The proposed method was critically analyzed based on distinct
perspectives and different formats to fulfill the basic criteria of image steganography. For the balance tradeoff,
we used different concepts or facets (i.e. magic matrix, MLEA, Key, LSB up to 4 bit, etc.) to cover all the criteria
of image steganography. Our proposed method ensures efficient Image format, dimension, and color that are
necessary for spatial, and frequency domains including Machine learning (ML) and Deep learning (DL) models.
The key contributions are given below.

• Herein, we design and propose a digital image steganography technique using the least significant bit substi-
tution.
• Our proposed technique incorporates reliable image steganography parameters and appropriate cover image
selection for achieving high security.
• Our proposed technique incorporates the random concepts for the selection of the cover image pixels by
inserting the message bits to outperform.
• The experimental results show that the proposed method is efficient as compared to the state-of-the-art meth-
ods.

The rest of the paper is partitioned into four divisions. Division II presents the connected exploration works,
Division III expounds on the proposed strategy, Division IV makes sense of the trial results and conversation
and Division V finishes up the research work.

Related work
Steganography using cover objects is a growing field of research that involves hiding secret information within
various objects. Recently, various methods of cover steganography have been developed, such as Most Significant
Bits and Least Significant Bits (LSB) (MSB-LSB), Pixel Value Differencing (PVD), etc. Respectively every
method has its pros and cons, subject to factors such as the implanting process of the hidden message, cover
medium, size of the secret data to stand encrypted, and other parameters7,8. However, despite ongoing efforts to
improve cover steganography, the error rates of these methods have not yet reached acceptable limits. Therefore,
evaluating existing research is crucial to making progress and improving the significance and effectiveness of
these methods. These limitations provide an opportunity for further advancement. This study focuses on LSB-
based cover steganography, which is a commonly used method for concealing confidential data within a cover
medium. Table 1 presents a security analysis of related works from the literature, Table 2 presents an acute
exploration of different image steganography approaches by assessing algorithms which are the basic criteria of
image steganography.
Tables 1 and 2 provide a detailed analysis of various research studies that focus on the security and quality
parameters of image steganography. The experimental results section of the studies discusses different concepts
for embedding secret messages within cover media using LSB-based methods. Additionally, the studies also
emphasize the quality of the stego images. Since there is a tradeoff among the steganography evaluation criteria,
some methods in the literature achieve good security but are unable to maintain good levels for other criteria
which are crucial for some efficient Stenographic methods. The effectiveness of image steganography depends
on the technique used to hide the secret message, as well as the choice of cover objects. Several different
approaches have been used to encrypt messages and hide them in cover mediums. In this context, two tables
have been created to critically analyze these methods based on quality assessment parameters and basic criteria.
The analysis aims to identify the advantages and disadvantages of each existing method. Comparatively, the
proposed algorithm has used a novel procedure of embedding the secret message and selecting the appropriate
object to tackle the aforementioned issues to improve image steganography.

Proposed stenographic technique

In this section, we explain our proposed technique. In our proposed technique, First, we define certain notations
that are utilized in the proposed algorithm. In Equations from 1 to 8, CI and FI represent the Cover image and
Flipped image, while the transposed image is denoted as a FT and SM denotes the stealthy message. In addition,
Skey and SIm denote the secret key and stego image used in the proposed method detailed explanation is used
in Figs. 3 and 4 and Algorithm 1. We will provide more details about the algorithm in the following sections.
We follow a series of steps to encode information into an image. First, we flip and transpose the cover image.
Subsequently, we divided the transposed image into Red, Green, and Blue channels by Eq.3. The Blue channel is
further divided into 4 equal blocks which are represented as BC1, BC2’, BC3’, and BC4’ respectively and finally,
the blue channel is shuffled by MGMx (Magic matrix). Then, the differencing values are calculated between the

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 3

www.nature.com/scientificreports/

Quality assessment metrics (QAM)

Techniques [Ref] Technique used MSE RMSE NCC SSIM PSNR
14 SCImage Steganography (IS) 0.789 0.999 0.998 1 50.04
15 Magic LSB IS 0.321 0.765 0.767 0.987 58.76
16 Secret Key 0.110 0.100 1 1 59.12
17 IS using DWT 0.999 0.977 0.789 0.899 61.11
18 Image S using DCT, OTP 0.001 0.112 0.989 0 65.23
19 IS based on Chaos theory 0.114 0.312 0.899 0.899 66.87
20 IS based on the Color Model 0.012 0.912 0.933 0.899 66.99
21 IS based histogram Analysis 0.963 0.678 0.923 0.987 68.98
22 IS based on Matching LSB 0.012 0.211 0 0.799 58.47
23 Pixel Indicator LSB based IS 0.889 0.999 1 0.788 65.36
24 GI and MLE based IS 0.889 0.889 0.989 0.887 69.01
25 IS using Modified LSB Approach 0.788 0.776 0.997 0.733 63.22
26 IS using an Adaptive pattern-based LSB
1 0.222 0.899 0.897 67.66
Approach
27 IS Approach for variability based on Flipping LSB 0.978 0.733 0.922 0.945 68.98
28 A Secret IS method using Map LSB 0.999 0.876 0.999 0.999 69.99

Table 1. Summary of some previous research works for image steganography using QAM.

Basic criterion for steganographic techniques

Technique used [Ref] Perception Security Temper protection Payload Computation
Ref14 Broke down quality up to some Confines Little 6, 8 KB randomization
Ref15 Not dependable Yes Reliable 6-10KB time consuming
Ref16 Only on RGB consistent Using key 10 kb No
Ref17 Not reliable trustworthy Consider as a low ratio Low payload Time-consuming
Ref18 Cover object some level commendable Trustworthy Squat Reasonable
Ref19 Low image quality Reasonable Can be considered reliable Low Reasonable
Ref20 High Reasonable Moderate High payload Time is high
Ref21 Acceptable limit High Easiest for attackers Low Low time
Ref22 Reasonable Low Sensible Great High
Ref23 Enough Sensible Low error ratio High Sensible
Ref24 Not high Sensible Modest Enough Sensible
Ref25 Due modification Enough Sensible Only up to 4 KB Reasonable
Ref26 Low Reasonable High Low level Not time consuming
Ref27 Workable High Acceptable ratio Low Reasonable
Ref28 practical High Reasonable Reasonable Reasonable
Ref29 Low Serviceable High Low Low
Ref30 High Serviceable Low Reasonable Low
Ref31 Workable Acceptable limit High Reasonable High

Table 2. Analysis of previous research works grounded on QAM’s.

secret message SM and the red channel by CalDiff. We then use a Multi-Level Encryption Algorithm (MLEA),
based on CalDiff and a secret key, Skey, to obtain the cipher text, CText. Using this function, we encode the cover
information into the shuffled Blue blocks using the secret key. Finally, we combine the three channels - Red,
Green, and Blue - and the secret key to achieve the stego-image, SIm. We use MLEA to embed the secret message
into the four shuffled blocks of the Blue channel, making it difficult for attackers to extract the information. The
magic matrix also adds a layer of security by shuffling the blocks of the Blue channel. The complete process is
illustrated in Fig. 5.

Embedding algorithm
Algorithm 1, First, we select the cover image and then flip and transpose it respectively. After dividing the FT
image into RGB channels, the Blue channel is again divided into 4 equal blocks for encrypting messages. In
addition, a magic matrix is used for shuffling the blue channel for security purposes. While red channel and
secret message bits are used for calculating the differing values. Now to generate cipher text, MLEA is applied on
bits using a secret key. Finally, using LSB the cipher text will be inserted into Blue channel sub 4 blocks. However,

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 4

www.nature.com/scientificreports/

Fig. 3. Proposed method.

Fig. 4. Mathematical Notations used in Proposed method (Eqs:1-8).

for shuffling the message bits we used magic matrix as a MATLAB function having some unique properties such
as rotation, reflection, etc., and having no repeated values as a result. The sum of this matrix row and column,
diagonal remains the same and this activity is placed for security purposes. For tamper and robustness, MLEA
and Key are utilized which are placed before the encrypted secret message to cover the image. MLEA has some
better operations which are usually applied before encoding secret messages into cover images. MLEA used
XOR operations for all bits at once which take 8 bits combination and replace 1st four bits with the last bits like
a flipping property. So the left circular shift operation is performed on every 8 bits and makes two equal blocks
array namely B1 and B2 in this way. IF B1 i = = 1, and XOR B2 by 1. Shown in Fig. 5.

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 5

www.nature.com/scientificreports/

Algorithm. 1. Embedding algorithm.

Fig. 5. Mlea for proposed stenographic method.

Extraction algorithm
Decoded message is the opposite of encoded by taking the stego image as input flipped and transposed respectively.
For getting the respective red, Green, and Blue channels set flag 1; and check the condition for getting the secret
message pixels from equals sub 4 blocks BC1’, BC2’, BC3’, and BC4’of blue channel. The complete steps of the
extraction algorithm are presented in Algorithm 2.

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 6

www.nature.com/scientificreports/

Algorithm. 2. EA (Extraction algorithm).

Fig. 6. Dataset.

Simulation and experimental results

To highlight the significance and importance of this research work, we have conducted a critical analysis of
some related works. After critical analysis the results in12 LSB sub-based15, , LSB-GLM21, LSB-RGB22, LSB-
Inverted based23,25,27, and Robust LSB28, compared with proposed method. A dataset of 165 different images was
downloaded from Image Processing Place (IPP) and the University of Southern California-Signal and Image
Processing Institute (USC-SIPI-2022) for proposed method testing32,33 shown in Fig. 6. The results of the tests
indicate that the proposed algorithm successfully embeds secret messages without any suspicious activity, noise,
or distortion. More details are provided below.

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 7

www.nature.com/scientificreports/

Fig. 7. Multiple viewpoints to analyze Stenographic algorithm’s performance.

Img’s M- Size Resulted PSNR

Baboon 81.32
Girl1 79.98
Tree 80.98
Lake 14KB 82.09
House 80.85
Peppers 78.00
Splash 86.06
Average on 165 images 81.32

Table 3. Persp-1 Images of size 512 × 512 pixels with 14 KB text.

Images Baboon Peppers Lake House Splash Flowers

128x 69.3 128x 74.2 128x 70.0 128x 71.2 128x 69.2 128x 77

Dimensions and 256x 77.1 256x 79.1 256x 75.2 256x 77.2 256x 73.2 256x 82.1
Resulted PSNR 512x 78.3 512x 81.3 512x 77.3 512x 82.3 512x 75.3 512x 83.3
1024x 81.3 1024x 83.9 1024x 85.9 1024x 84.9 1024x 80.9 1024x 87.9
Average resulted PSNR on different dimension: D- dimensions
Average of 165 images
D-128 = 71.81, D-256 = 77.31, D-512 = 79.63, and D-1024 = 84.133

Table 4. Persp-2: Resulted PSNR values on different images using P2.

Performance
The proposed technique is assessed in light of distinct viewpoints with some color images to show the significance
and inspiration of this work. So, the basic assessment parameter for analyzing both cover and Stego images is
the Peak Signal Noise Ratio (PSNR). PSNR value determines both image quality. If PSNR is > 30dB, the image is
high-quality. Equation (9) finds PSNR value34. Different analysis perspectives are given in Fig. 7.
( )
Cmax 2
PSNR = 10log10 (9)
MSE

The algorithm underwent thorough analysis, and the results are in Tables 3, 4, 5 and 6 respectively.
The results obtained from different angles demonstrate the superior quality of the proposed algorithm
in comparison to existing strategies. A crucial aspect to consider is determining the optimal size of message
bits embedded in an image of a particular size to ensure efficient steganography. Hence, before embedding
the secret message into the image it is important to compute the message and image pixels to point out the
best dimension and format images for which size of the text. Consequently, the proposed strategy is assessed

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 8

www.nature.com/scientificreports/

Images
512 × 512 Baboon Peppers Lake House Splash Flowers
8KB 77.2 8KB 69.0 8KB 65.1 8KB 70.1 8KB 89.0 8KB 75.3

Different Size of Text and 10KB 79.1 10KB 72.1 10KB 69.2 10KB 73.2 10KB 85.2 10KB 81.3
Resulted PSNR 12KB 80.1 12KB 77.3 12KB 70.2 12KB 74.4 12KB 83.0 12KB 76.2
14KB 82.2 14KB 81.9 14KB 76.2 14KB 79.6 14KB 77.9 14KB 78.1
Average resulted PSNR on different size of text: ST (Size text)
Average of 120 images
ST-8 = 74.28, ST-10 = 76.68, ST-12 = 7686, and ST-14 = 79.81

Table 5. Persp-3: Resulted PSNR on 512 × 512 embedded 14, 10, 12, and 8 KB size of text.

PSNR values
Images Message Size PNG TIFF BMP JPG
Baboon 79.321 73.123 69.212 82.212
Girl 81.223 77.765 80.121 82.918
Peppers 86.189 71.554 68.876 82.887
Lake 14KB 83.098 79.986 75.097 82.643
House 80.854 79.654 87.001 80.087
Flowers 84.001 87.009 87.091 85.098
Splash 86.065 87.112 83.087 86.076
An average of 150 images 82.82 79.33 78.78 83.00

Table 6. Persp-4: Resulted PSNR on (PNG, BMP, JPG, D-512 × 512) using 14 KB text.

Comparison with some existing methods

Image Baboon Flowers Peppers Lake House Tree Splash
GLM 21 78.34 79.09 74.00 79.86 69.00 77.32 80.22
Secure RGB
22 77.99 81.00 75.00 81.31 78.98 82.10 80.00

Inverted
70.91 88.00 71.98 79.99 79.00 80.00 82.07
LSB 23
LSB IMST 24 73.04 80.00 80.32 82.98 83.09 84.32 83.32
IMST 27 76.99 84.02 83.87 80.09 80.87 82.98 82.97
Robust ST 28 77.32 78.43 79.22 76.66 78.32 79.97 79.97
Proposed
79.34 87.98 85.01 84.01 83.89 84.98 88.12
method

Table 7. PSNR values-based state-of-the-art methods with proposed method.

utilizing different viewpoints. It is also important as a principle perspective to distinguish which cover object is
the most appropriate for encrypting the secret messages to pledge secure transmission over the web. To solve this
issue, that’s why the proposed work has tried to utilize different image formats, including PNG, JPG, and BMP,
dimensions, and viewpoints in light of PSNR values to show the method performance. Using various QAMs the
proposed work is comparatively analyzed for security purposes in the following section.

Analysis of the proposed steganography technique

To show the viability and significance of the proposed work against different facts, such as trimming and scaling,
it was tested using several quality assessment metrics outlined in this section. The results indicate significant
improvements and better performance of the proposed algorithm.

Performance analysis
This section provides a detailed explanation of the results obtained from the suggested algorithm. These results
are also compared with previous research studies based on metrics such as MSE, RMSE, NCC, SSIM, and PSNR.
Table 7; Fig. 8 present the experimental outcomes of the proposed technique based on these assessment metrics,
demonstrating the enhancements made by this method. Here are some important assessment metrics that
need to be elaborated for the experimental results of any steganographic method. So, to measure the difference
between both input and resulting images, Root Mean Square Error (RMSE) is usually used. One more significant
quality metric that is utilized to examine the connection between an input and results from media is Normalized
Cross Correlation (NCC). If the value of the NCC is equal to 1, then both cover and stego images are considered

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 9

www.nature.com/scientificreports/

Fig. 8. Comparison with State of the art models.

identical. On the other hand, if this value becomes 0, it indicates that there is a complete difference between both
images. The Structural Similarity Index (SSIM) is a significant quality metric used to evaluate the quality of cover
and stego images. It consists of three parts followed by Luminance, Contrast, and Structural and it determines
the quality of the image. An image will be considered a quality image if the value of SSIM is 1 while 1 < presents
the difference between the cover and stego image. Equation 13 elaborates on the formulas for each segment. If
measuring the contrast between both cover and stego image Mean Squared Error (MSE) is usually used and
both images will be considered if the value of MSE is 0. To obtain quality and robust images MSE value should
be low35,36 Formulas of each are elaborated in Eqs. 10–13.
√( )
∑ 1
RMSE = x=1 (Cx
N
− Sx )2 (10)
N
√( )
∑ 1
RMSE = x=1 (Cx
N
− Sx )2 (11)
N
∑M ∑ N
y=1 (S (x, y) *C (x, y))
NCC = x=1
∑M ∑N 2
(12)
x=1 y=1 S (x, y)
( )
2µ x µ y + C1 (2σ xy + C2 )
SSIM (X, Y) = ( ) (13)
µ x 2 + µ y 2 + C1 (σ x 2 + σ y 2 + C2 )

This section elaborated on the results of the proposed method using some security measures and QAMs to
ensure the work’s importance and resistance. Before continuing to the outcomes, making sense of these concepts
is significant and explained below.
The Relationship Coefficient (CC) is utilized to track down the linearity (degree and heading) of two irregular
factors. It assumes an imperative part because of its estimation qualities37. If the CC value is equal to 1, the two
factors are supposed to be somewhat similar or firmly related. Equation 14 shows the formula for CC.
∑
(xi − xm ) (yi − ym )
I = ∑ √∑
i
√∑ (14)
i i (xi − xm )2 i (yi − ym )2

Image fidelity (IF) is also a more significant metric that tracks down the picture quality. Equation 15 shows the
formula for IF calculation where P and S address the Cover and Stego picture, and “I” and “j” address beginning
and finishing values.
∑
i,j (P (i, j) − S (i, j))2
IF = 1 − ∑ (15)
i,j (P (i, j) × S (i, j))

To measure the quality of stego images Quality Index is also used on different perspective. Equation 16 elaborates
the detailed formula that how to calculate the stego images. T and H represent both stego and cover images and
n shows the pixel quantity in an image while the value scope of QI is in the range of -1 to 1. Both input and
resulting images will be identical if the value is = 1, otherwise, the images will show the difference between the
images.
4σ HT H′ T ′
Q= ( )
σ 2H+ σ 2T (H ′ 2 + T ′ 2 )

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 10

www.nature.com/scientificreports/

Resulted Values
Image Correlation coefficient IF QI
Lake 0.999 0.999 1
Flowers 0.999 1 1
Peppers 1 1 0.999
Baboon 0.998 0.998 1
Splash 0.999 0.999 1
Tree 0.999 0.999 0.998
House 0.998 1 1

Table 8. Cumulative QAM results of the Proposed Method.

Effect on red, green, and blue statistical analysis

Analysis Parameter Images Cover Image Stego Image
Red Green Blue Red Green Blue
Baboon 2.666 1e + 03 2.4447e + 03 1.23478e + 03 2.666 1e + 03 2.4447e + 03 1.23478e + 03
Contrast Girl 7.2324e + 08 8.2721e + 05 7.3514e + 08 7.2324e + 08 8.2721e + 05 7.3514e + 08
Peppers 2.5134e + 08 2.7042e + 08 8.1540e + 06 2.5134e + 08 2.7042e + 08 8.1440e + 07
Baboon 1.2334e + 06 1.3234 + 07 2.7244e + 05 1.2334e + 06 1.3234 + 07 2.7244e + 05
Homogeneity Girl 7.3224e + 07 7.2440e + 08 5.4152e + 08 7.3314e + 07 7.2440e + 08 5.4152e + 08
Peppers 1.7453e + 07 1.3462e + 07 2.3638e + 09 1.7453e + 07 1.4462e + 07 2.3638e + 09
Baboon 6.6632 5.4632 6.5345 6.6632 5.4632 6.5345
Entropy Girl 8.5857 8.6634 5.1457 8.5857 8.6634 5.1457
Peppers 7.4865 9.3251 7.3735 7.4865 9.3251 7.3735

Table 9. Effect analysis of red, green, and blue channels while embedding message bits.

1 ∑ ( )2
σ 2H = N
Hi − H ′
N −1
i=1

1 ∑ 1 ∑
H′ = N
i=1 Hi −T = N
i=1 Ti
N N
1 ∑ ( )2
σ 2H = N
Ti − T ′ (16)
N −1
i=1

Contrast analysis is widely used to quantify the intensity dissimilarity between pixels and their neighbors
throughout the image. It also helps in the main texture area of the image and can be calculated as under in eq. 17.
4σ HT H′ T ′
Q= ( ) (17)
σ 2 2
H+ σ T (H ′ 2 + T ′ 2 )

To measure the security of any steganographic method Information Entropy (IE) is usually used. Suppose e1….
em, as m shows the measure of potential components and from P(e)… P(em) will be considered as a probability
for thesis components. The formula in eq. 18 elaborated on the calculation of Entropy. In view of the recurrence
of the image, the condition is the assessment of the typical number of pieces expected to implant a series of
pieces.
1 ∑ ( )2
σ 2H = N
Hi − H ′ (18)
N −1
i=1

For gray level diagonal and co-occurrences, Homogeneity (H) is widely used to find the closest level of element
distribution38,39. Eq.19 shows the whole calculation. So i,j show lists of line and segment numbers, and p (i, j)
represent pixel’s values ith column and jth section.
1 ∑ 1 ∑
H′ = N
i=1 Hi −T = i=1 Ti (19)
N
N N

The proposed calculation has been demonstrated critical in light of observational outcomes from QAMs.
The investigation depends on differentiation, homogeneity, and entropy to recognize the specific distinctions
between the embedded medium’s relating channels (red, green, blue) after embedding the cipher message shown
in Tables 8 and 9.

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 11

www.nature.com/scientificreports/

The results of our proposed method were better than those of the techniques described in27,29. According to
Table 10, our method has a relatively high hiding capacity compared to existing techniques, with a normal PSNR.
The average bpp value of our method for more than eight test images is 5.025.

Security analysis
The section expounded on the security of the proposed technique has been talked about concerning Pixel
Difference Histogram (PDH), and RS steganalysis. The PDH and RS analysis is the normal and powerful
measurable steganalysis to recognize the presence of the restricted data in the stego-picture for the LSB based
techniques40. The LSB replacement methods experience the ill effects of RS investigation and PVD methods
experience the ill effects of PDH investigation. The proposed method utilizes ideas like adjusted LSB replacement,
also value differencing, so both RS investigation and PDH should dissect it. Here, RS steganalysis is used to break
down the effectiveness of the extended procedure41. Consequently, the RS plot applied more than 500 implanted
pictures to check or hack the secret-covered picture. So, both RS and PDH considerations is proceeded according
to the structure talked about in42,43. RS analysis of two standard images Lena, baboon, Pepper, and house are
presented in Fig. 9a-c respectively. The x-axis and y-axis describe the ratio of the concealing capability and
ratio of the singular and regular groups. To plot four arcs/curves, four factors S-m, Sm, R-m, Rm as described
in41,44,45. RS analysis identifies the presence of implanted bits based on the condition if Rm-Sm < R-2m-S. While
the steganographic-based method is unnoticeable by RS analysis if Sm ≈ S− m < R m ≈ R− m.
Figure 10a, b indicate that the condition Sm ≈ S − m < Rm ≈ R − m holds because the curves of R − m and Rm
are positioned above the curves of R − m and Rm, while the curves of S − m and Sm are equal in length to each
other. Additionally, the curves of R − m and Rm are also of equal length. This leads to the conclusion that the
intended method is highly resistant in terms of RS analysis. To validate the effectiveness of the proposed method,
we need to use the concepts of LSB and value differencing to examine its success through PDH analysis and RS
analysis. To show the difference on both axis’s for frequency and pixel difference PDH graph is used. The PDH
analysis of some standard images namely Lena, pepper images presented in Fig. 10a,b respectively. The dotted
and solid lines indicate the PDH analysis of both stego and cover image respectively. It very well may be seen
that the step impact or crisscross presence is tiniest. While the arcs/curve show the smooth nature of the stego
image. Hence, it very well may be presumed that the proposed strategy is pitifully impervious to PDH check.

Proposed method histogram analysis

The proposed method also conducted histogram analysis using some images. As we know that histogram is
utilized to decide the specific frequencies of every pixel in the picture and to uncover point-by-point contrasts
between the cover and stego pictures. We fundamentally assess the proposed technique by breaking down the
histograms of a few pictures, including Strawberry, Peppers, Mandrill, and House. The outcomes exhibit the
viability of the proposed work, as shown in Fig. 1146.

Conclusion and future work

In this research, we propose a secret key-based LSB substitution image steganography technique. Our proposed
technique incorporates reliable image steganography parameters and appropriate cover image selection
for achieving high security. Our proposed method utilized some random concepts for the selection of the
cover image pixels for inserting the message bits to get outperformance. To demonstrate the effectiveness
and enactment, we performed experimentation. The experimental results show that the proposed method is
critically analyzed based on various steganography and different statistical metrics. The achieved results show
our proposed method performs in terms of various QAMs such as MSE, RMSE, SSIM, NCC, PSNR, correlation
coefficient, image fidelity, quality index, contrast, entropy, and homogeneity. Moreover, it has been confirmed
that the given method weakly endures PDH analysis and very strongly endures RS analysis. In addition, the
empirical results show a high level of robustness, acceptable limits of payload, imperceptibility, and resistance
against attacks scaling, cropping, tempering, etc. Time complexity, limited payload, and RGB color model are
demerits of the work. Combining image steganography with cryptography to achieve a better steganography
scheme, may be investigated as future work. In the spatial domain, compression methods such as Huffman code,

Comparison results of the proposed and existing methods

Kamil, S 27 Ray 25 Proposed Method
Image
512 × 512 bpp Hiding Capacity PSNR bpp Hiding Capacity PSNR bpp Hiding Capacity PSNR
Baboon 928,000 3.21 43.32 881,383 3.01 44.59 1,152,683 4.21 56.12
Girl 821,344 3.22 43.18 851,343 3.22 44.21 1,032,421 4.12 55.01
Peppers 946,722 3.11 43.18 851,348 3.12 44.33 1,187,381 4.21 51.32
Lake 884,783 3.82 44.18 811,211 2.92 44.33 8,513,432 4.12 52.11
House 883,247 3.21 43.18 741,311 2.89 44.23 8,513,483 4.12 50.31
Tree 928,374 3.41 44.18 844,665 2.76 43.88 1,188,778 4.22 50.11
Lena 851,343 3.51 43.38 854,654 3.20 44.89 1,109,887 4.22 59.11
Splash 851,348 3.51 43.38 745,334 2.98 44.01 1,176,767 4.22 50.32
Average 886,895 3.37 43.49 764,138 3.01 44.30 12,459,962 4.31 52.17

Table 10. Comparison of proposed method with state-of-the-art methods.

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 12

www.nature.com/scientificreports/

Fig. 9. RS Plot of Lena, pepper, Baboon, and House standard images.

Fig. 10. PDH Plot of Two standard images namely Lena and Pepper.

HSI model, and acronoment method are also best for future work. Also, some machine learning techniques
such as different deep learning architecture or unsupervised learning may be utilized for obtaining efficient
steganography for secure communication between the sender and the receiver. Now GAN GAN-based models
are also can be used for better and secure communication.

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 13

www.nature.com/scientificreports/

Fig. 11. Histogram analysis of various cover and stego images.

Data availability
Data is provided within the manuscript or supplementary information files.

Received: 7 October 2024; Accepted: 11 December 2024

References
1. Subramanian, N., Elharrouss, O., Al-Maadeed, S. & Bouridane, A. Image steganography: A review of the recent advances. IEEE
access. 9, 23409–23423 (2023).
2. Hemeida, F., Alexan, W. & Mamdouh, S. A comparative study of audio steganography schemes. Int. J. Comput. Digit. Syst. 10,
555–562 (2021).
3. Lakshmi Sirisha, B. & Chandra Mohan, B. Review on spatial domain image steganography techniques. J. Discrete Math. Sci. Crypt.
24 (6), 1873–1883 (2021).
4. Singh, K. U. A survey on image steganography techniques. Int. J. Comput. Appl., 97(18). (2014).
5. Menon, N. A survey on image steganography. In 2017 International Conference on Technological Advancements in Power and
Energy (TAP Energy) (pp. 1–5). IEEE. (2017), December.
6. Kaur, S., Singh, S., Kaur, M. & Lee, H. N. A Systematic Review of Computational Image Steganography Approaches. Arch. Comput.
Methods Eng., 1–23. (2022).
7. Ejidokun, T., Omitola, O. O., Nnamah, I. & Adeniji, K. Implementation and Comparative Analysis of Variants of LSB Stenographic
Method. In 2022 30th Southern African Universities Power Engineering Conference (SAUPEC) (pp. 1–4). IEEE. (2022), January.
8. Ansari, A. S., Mohammadi, M. S. & Parvez, M. T. A multiple-format steganography algorithm for color images. IEEE Access. 8,
83926–83939 (2020).
9. Bhavani, Y., Kamakshi, P., Kavya Sri, E. & Sindhu Sai, Y. A Survey on Image Steganography Techniques Using Least Significant Bit.
In Intelligent Data Communication Technologies and Internet of Things (281–290). Springer, Singapore. (2022).
10. Sahu, A. K. & Gutub, A. Improving grayscale steganography to protect personal information disclosure within hotel services.
Multimedia Tools Appl., 1–21. (2022).
11. Hussain, M. & Hussain, M. A survey of image steganography techniques. (2013).
12. Mohamed, M. M., Ghoniemy, S. & Ghali, N. I. A Survey on Image Data Hiding Techniques. Int. J. Intell. Comput. Inform. Sci., 1–25.
(2022).
13. Nabi, S. T., Kumar, M., Singh, P., Aggarwal, N. & Kumar, K. A comprehensive survey of image and video forgery techniques:
variants, challenges, and future directions. Multimedia Syst., 1–54. (2022).
14. Muhammad, K., Ahmad, J., Rehman, N. U., Jan, Z. & Qureshi, R. J. A secure cyclic Stenographic technique for color images using
randomization. arXiv preprint arXiv:1502.07808. (2015).
15. Alanzy, M., Alomrani, R., Alqarni, B. & Almutairi, S. Image steganography using LSB and hybrid encryption algorithms. Appl. Sci.
13 (21), 11771 (2023).
16. Karim, S. M., Rahman, M. S. & Hossain, M. I. A new approach for LSB based image steganography using secret key. In 14th
international conference on computer and information technology (ICCIT 2011) (pp. 286–291). IEEE. (2011), December.

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 14

www.nature.com/scientificreports/

17. Baby, D., Thomas, J., Augustine, G., George, E. & Michael, N. R. A novel DWT based image securing method using steganography.
Procedia Comput. Sci. 46, 612–618 (2015).
18. Rachmawanto, E. H. & Sari, C. A. Secure image steganography algorithm based on dct with otp encryption. J. Appl. Intell. Syst. 2
(1), 1–11 (2017).
19. Naima, S. et al. Secure and imperceptible frequency-based watermarking for medical images. Circuits, Systems, and Signal
Processing, 1–22. (2024).
20. Çataltaş, Ö. & Tütüncü, K. Comparison of LSB image steganography technique in different color spaces. In 2017 international
artificial intelligence and data processing symposium (IDAP) (pp. 1–6). IEEE. (2017), September.
21. Huang, R., Lian, C., Dai, Z., Li, Z. & Ma, Z. A novel hybrid image synthesis-mapping framework for steganography without
embedding. IEEE Access. (2023).
22. Dhar, S. & Sahu, A. K. Digital to quantum watermarking: A journey from past to present and into the future. Comput. Sci. Rev. 54,
100679 (2024).
23. Dhar, P. K., Kaium, A. & Shimamura, T. Image steganography based on modified lsb substitution method and data mapping. Int.
J. Comput. Sci. Netw. Secur. 18 (3), 155–160 (2018).
24. Gutub, A. A. A. Pixel indicator technique for RGB image steganography. J. Emerg. Technol. web Intell. 2 (1), 56–64 (2010).
25. Ray, A. M., Pramanik, S., Das, B. & Khanna, A. Hybrid cryptography and steganography method to provide safe data transmission
in IoT. In International Conference on Data Analytics & Management (pp. 513–524). Singapore: Springer Nature Singapore., June.
(2023).
26. Sahu, A. K. & Sahu, M. Digital image steganography and steganalysis: A journey of the past three decades. Open. Comput. Sci. 10
(1), 296–342 (2020).
27. Sharma, D. & Prabha, C. Hybrid security of EMI using edge-based steganography and three-layered cryptography. In Appl, Data
Sci. Smart Syst. (pp. 278–290 ). CRC.
28. Venkata Krishna, G. P. C. & Vivekananda Reddy, D. Machine learning-enhanced hybrid cryptography and image steganography
algorithm for securing cloud data. J. Intell. Fuzzy Syst., (Preprint), 1–11. (2024).
29. Almazaydeh, L. Secure RGB image steganography based on modified LSB substitution. Int. J. Embed. Syst. 12 (4), 453–457 (2020).
30. Rustad, S., Syukur, A. & Andono, P. N. Inverted LSB image steganography using adaptive pattern to improve imperceptibility.
Journal of King Saud University-Computer and Information Sciences. (2021).
31. Kamil, S., Abdullah, S. N. H. S., Hasan, M. K. & Bohani, F. A. Enhanced flipping technique to reduce variability in image
steganography. IEEE Access. 9, 168981–168998 (2021).
32. ALabaichi, A., Al-Dabbas, M. A., A. A., K. & Salih, A. Image steganography using least significant bit and secret map techniques.
International journal of electrical & computer engineering (2088–8708), 10(1). (2020).
33. Liu, X. et al. Robust coverless steganography using limited mapping images. J. King Saud University-Computer and Information Sci..
(2022).
34. Naz, M. T. S. A. & Zade, S. A new approach for image steganography using (Inter Pixel Value Difference and Quantized Range Table
Method, 2022).
35. Xu, Y., Mou, C., Hu, Y., Xie, J. & Zhang, J. Robust invertible image steganography. In Proceedings of the IEEE/CVF Conference on
Computer Vision and Pattern Recognition (pp. 7875–7884). (2022).
36. https://sipi.usc.edu/database
37. https:​​​//w​ww.imageprocessingpl​ace​.com​/root​_f​i​ les_​V3/image_​databases.htm
38. Alatiyyat, B. F. & Narmatha, C. Survey on image steganography techniques. In 2022 2nd International Conference on Computing
and Information Technology (ICCIT) (pp. 57–64). IEEE, January. (2022).
39. Nabi, S. T., Kumar, M., Singh, P., Aggarwal, N. & Kumar, K. A comprehensive survey of image and video forgery techniques:
variants, challenges, and future directions. Multimedia Syst. 1, 54 (2022).
40. Wang, Z., Zhou, M., Liu, B. & Li, T. Deep image steganography using transformer and recursive permutation. Entropy 24 (7), 878
(2022).
41. Sharif, A., Mollaeefar, M. & Nazari, M. A novel method for digital image steganography based on a new three dimensional chaotic
map. Multimedia Tools Appl. 76, 7849–7867 (2017).
42. Zhang, W. et al. Continual learning for blind image quality assessment. IEEE Trans. Pattern Anal. Mach. Intell. (2022).
43. Swain, G. Two new steganography techniques based on quotient value differencing with addition-subtraction logic and PVD with
modulus function. Optik 180, 807–823 (2019).
44. Pradhan, A., Sahu, A. K., Swain, G. & Sekhar, K. R. Performance evaluation parameters of image steganography techniques. In
2016 International Conference on Research Advances in Integrated Navigation Systems (RAINS) (pp. 1–8). IEEE. (2016), May.
45. Sahu, M., Padhy, N., Gantayat, S. S. & Sahu, A. K. Performance analysis of various image steganography techniques. In 2022 Second
International Conference on Computer Science, Engineering and Applications (ICCSEA) (pp. 1–6). IEEE. (2022), September.
46. Mileva, A., Velinov, A., Dimitrova, V., Caviglione, L. & Wendzel, S. Information hiding in the DICOM message service and upper
layer service with entropy-based detection. Entropy 24 (2), 176 (2022).

Acknowledgements
This research is funded by the European University of Atlantic.

Author contributions
“Shahid Rahman.Jamal uddin. and Farhan Amin.Hameed Hussain. wrote the main manuscript text and Sabir
Shah.Abdu Salam. Isabel de la Torre Díez. Debora Libertad Ramírez Vargas and Julio César Martínez Espinosa
prepared Figs. 1-3. All authors reviewed the manuscript.”

Declarations

Competing interests
The authors declare no competing interests.

Additional information
Correspondence and requests for materials should be addressed to F.A. or I.T.D.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 15

www.nature.com/scientificreports/

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives

4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in
any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide
a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have
permission under this licence to share adapted material derived from this article or parts of it. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated
otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence
and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to
obtain permission directly from the copyright holder. To view a copy of this licence, visit ​h​t​t​p​:​//​ ​c​r​e​a​t​i​v​e​co
​ ​m​m​o​
n​s​.​o​r​g/​ ​l​i​c​e​n​s​e​s​/b
​ ​y​-​n​c​-​n​d​/​4​.​0​/​​​​.​ ​​

© The Author(s) 2024

Scientific Reports | (2025) 15:107 | https://doi.org/10.1038/s41598-024-83147-3 16