A Huffman code LSB based image steganography technique using multi-level encryption and achromatic component of an image

In the recent couple of years, due to the accelerated popularity of the internet, various organizations such as government offices, military, private companies, etc. use different transferring methods for exchanging their information. The Internet has various benefits and some demerits, but the primary bad mark is security of information transmission over an unreliable network, and widely uses of images. So, Steganography is the state of the art of implanting a message in the cover objects, that nobody can suspect or identify it. Therefore, in the field of cover steganography, it is very critical to track down a mechanism for concealing data by utilizing different blends of compression strategies. Amplifying the payload limit, and robustness, and working on the visual quality are the vital factors of this research to make a reliable mechanism. Different cover steganography research strategies have been recommended, and each adores its benefits and impediments but there is a need to foster some better cover steganography implements to accomplish dependability between the essential model of cover steganography. To handle these issues, in this paper we proposed a method in view of Huffman code, Least Significant Bits (LSB) based cover steganography utilizing Multi-Level Encryption (MLE) and colorless part (HC-LSBIS-MLE-AC) of the picture. It also used different substitution and flicking concepts, MLE, Magic matrix, and achromatic concepts for proving the proficiency, and significance of the method. The algorithm was also statistically investigated based on some Statistical Assessment Metrics (SAM) such as Mean Square Error (MSE), Peak Signal Noise Ratio (PSNR), Normalized Cross Correlation (NCC), Structural Similarity Index Metric (SSIM), etc. and different perspectives. The observational outcomes show the likelihood of the proposed algorithm and the capacity to give unwavering quality between security, payload, perception, computation, and temper protection.

For data transmission, the Internet has become an excellent system, due to its accelerated popularity, inexpensiveness, and efficiency of it[1]. Therefore, for sharing information in digital form over the Internet, one file image is widely used and becomes very easy to send over the Internet[2]. In addition, it is very easy to copy or modify the transmitted data over the internet by unauthorized persons or attackers. There are various identified tools available to make the exploitation of secret information, security, privacy, etc. being transmitted and also to make the possibility of different vulnerabilities, attacks, and hateful threats such are scaling, cropping, tempering, spoofing, phishing, eavesdropping, privilege escalation, clickjacking, social engineering, bot, backdoor, viruses, botnet, malware and many more[3],[4].To deal with secure correspondence over the Internet, various techniques are proposed in the recent couple of years but each has related pros and cons. Thus, to satisfy the requirement for secure correspondence, it is important to make superior ways of making a safe framework to satisfy the requirement for transmission over the web between users. However, the ancient method used for secure communication in order to provide a safe way of transferring data over the internet is Encryption[5]–[7]. It converts normal text means plain text to cipher text using some keys. So, encryption is a way of converting plain text into cipher text within any cover media used for encryption, that there is no clue of the existing data and no one can suspect the encrypted information. Furthermore, the recipient might involve a mystery key for encryption and just the key supervisors can unscramble the mystery message utilizing the key which is to be given by the source. Using the encryption concepts there are some multimedia information mediums used such are image, text, audio, video, network, etc.[8].

Therefore, the main idea of encryption is the notion of steganography which hides secret information within the cover objects without knowing the existence of the information within it. Steganography is the part of data concealing that encodes the mystery message that nobody can identify it. Image steganography shows an essential part of protected communication in this digital world because one file type that is rapidly used is the image[9]. Furthermore, image steganography uses an image as a cover object and inserts the secret information using different reported methods such as spatial domain, frequency domains, etc. The extensive surveys of image steganography are presented in various works as described later in "Summary of the related works"[10]–[14]. Every existing method has its own advantages and disadvantages in terms of payload, security, perception, temper protection, and computation which are the basic criteria of steganography, as well as, the basic needs for any steganographic methods as shown in Fig. 1[6].

Graph: Figure 1The need for image steganography techniques.

Though, we have proposed an improved and novel technique based on Multi-Level Encryption (MLE), an achromatic component of an image, and Huffman LSB. The proposed method also used some transposition, and magic matrix concepts to expand the significance and inspiration of the proposed technique. For embedding the secret message, the proposed algorithm used the Huffman coding priority concept and made the Huffman encrypted sequences, and then embedded them into the cover media. After making the Huffman encrypted sequences, different encrypted operations applied to it made the method outperform others. This is because the proposed algorithm used an embedded procedure of the cover data within the image in a manner that if anyone suspects the stego image or extracts the message, then they will not extract the encoded message version. Only the holders of the Huffman table or tree can regain the secret message correctly because the communication is between the two parties. Therefore, the proposed research work has a novel and improved contribution in terms of security, transparency, payload, and temper protection in order to make it significant. Some main objectives and contributions of the said method are given as follows.

• A proposed technique that uses the Huffman coder priority for encrypting the secret message and the result is the form of Huffman encrypted sequences to improve the reliability between basic criteria such as temper protection and security.
• Achromatic components of an image Hue Saturation Intensity (HSI) variety model are utilized rather than the RGB pictures variety model to decrease the handling time and increment the security.
• For making deciphering the mystery message testing, and thought-provoking, the proposed method divided the I channel or plane (I-Plane) into four equal blocks and shuffled the blocks using Magic Matrix, and the method also used MLE for giving the tough time to attackers and to increase the security.
• The proposed strategy was fundamentally dissected in the opinion of various key view-points (i.e. different sizes of images with different sizes of text, different images with the same text size, and different format images) to work on the proficiency and viability of the scheme.

In sum-up, the basic criterion of cover steganography is presented in Fig. 1, which elaborates the payload maximum amount of furtive data to be implanted within the cover object. Transparency shows the quality of the image. Furthermore, robustness spectacles the retreat of the stego image, that the stego object is undamaged after embedding even if attacked. shows the opposition against various assaults of the cover object. While computation explains the time intricacy of the implanting system into the cover object[15]. The rest of the paper is organized into various sub-sections as follows. Section 2 elaborates on the abstract details of the related reported research works. Section 3 explains the proposed methodology and algorithms. Sections 4 describes experimental parameters and obtained results. Finally, Sect. 5 concludes this work along with directions for future work.

For secure communication between sender and receiver in this digital world over the internet, cover steganography plays a vital role and flourishing research area. For digital steganography, many research works have been proposed over the last decades such as LSB, pixel value differencing, randomization, cyclic, etc. and each has their related pros and cons. Notably, for inserting the mystery message inside the cover media the most well-known and habitually utilized strategy is the LSB technique[5]. The wide use of this technique is due to its simplicity and straightforwardness. Therefore, this section elaborates on the basic concepts of RGB and HSI color model, Least Significant Bit (LSB), and critical analysis of some methods presented in the literature[16]. However, how about we make sense of RGB and HSI variety models; one of the main parts of any item is its tone.

The utilization of variety in picture handling is persuaded by standard two aspects.

• From a scene, that often streamlines object identification and abstraction, a color is a powerful descriptor.
• Contrasted with about just too many shades of gray the people can discern thousands of color shades and intensities.

The second is predominantly significant in physical image analysis (i.e. when performed by people). The purpose of a color model or color system is to facilitate the specification of colors in some standard, and generally accepted way when working in image processing because it is the main part of it. Each color tone is addressed by a solitary point because color mode is a description of a coordinate system and subspace within that system. In practice most commonly used color models; Red, Green, Blue (RGB) model used for monitors and a broad class of color video cameras because it is the hardware-oriented models in terms of digital image processing. For printing the second color model used is Cyan, Magenta, Yellow and Black (CMYK), Cyan, Magenta, Yellow (CMY). To correspond closely with the way humans, describe and interpret color, the third color model Hue, Saturation, Intensity (HSI) is used. So, the main focus of this study is the HSI model. In HSI color Model, changing from one model to the other is an open process while creating colors in the RGB and CMY models[17]. For hardware implementations, this ideally suited this color system. The RGB framework coordinates pleasantly with the way that the natural eye is firmly viewpoint to red, green, and blue primaries. Yet, the RGB, CMY, and other comparable variety models are not appropriate for depicting colors in wording that are pragmatic for human translation.

For instance, one doesn't allude to the shade of color by giving the level of every one of the primaries creating its tone. Besides, we don't consider a variety of pictures as being made out of three essential pictures that join to frame that solitary picture[18]. So, we describe by its hue, saturation, and brightness, when humans view a color object. Hue is a color aspect that defines a clean color (i.e. Yellow, red, and orange), while saturation gives a measure of the degree to which a pure color is diluted by a white light. Whereas intensity is a particular descriptor that is basically difficult to gauge. It is one of the key aspects in describing the color sensation, and it also exemplifies the colorless thought of intensity. This quantity absolutely is measurable and simply interpretable because it is the utmost convenient descriptor of monochromatic images. In a color image, the HSI model decouples the intensity component from the color-carrying information (hue and saturation). Based on color descriptions HSI color model is an ideal tool for developing image processing algorithms because they are naturally intuitive to humans. For color generation, we can say that RGB is an ideal tool, but for color description, it is significantly more restricted. Hereafter, from the RGB image, we should be capable to excerpt intensity[19],[20]. So, the HSI color model plays a vital role in secret information camouflage because the Intensity plane (I Plane) does not grieve the other planes, unique RGB diverts in which all planes are unequivocally co-related with one another. Besides, handling an image in the HSI model is somewhat more economical based on LSB-based techniques, etc. due to its unique properties. So, LSB is the process of embedding the secret message into cover image pixels of LSBs either randomly or sequentially shown in Fig. 2[3],[9],[21]. The given figure explores the concept of LSB by taking the cover image converting it into its corresponding American Standard Code Information Interchange (ASCII) values then converting it into binary values[22]. After reading the image pixels' values and their binary forms it converts the secret message values to binary form[23]. Furthermore, it is the procedure of implanting secret data bit K replaced with cover image pixels K LSB. Equation (1) represents the embedding process.

1 $${\text{S}}_{(\text{i},\text{j})}^{\text{IM}}={\text{C}}_{(\text{i},\text{j})}^{\text{I}}-{\text{C}}_{(\text{i},\text{j})}^{\text{I}}{\text{mod}}_2^{\text{K}}+{\text{S}}_{(\text{i},\text{j})}^{\text{M}}$$

Graph

Graph: Figure 2The basic concept of least significant bit (LSB).

Where SIM (i, j) presents the stego image i, where j represents row and column, CI is the cover image, and SM secret message. Different LSB-based methods are proposed recently but they still have issues to improve this research area. The human eye is very naked and has the properties to detect or suspect little change in any smooth area of the images[24]. So, encrypting the secret message with the image either randomly or sequentially, and not all pixel values are used for embedding so the change may be happening on only the embedding area of the image. Therefore, various image steganographic reported works are adapted to embed a suitable amount of message in appropriate cover objects[25].

Lee et al. recommended a high implanting message-based image steganography where they embedded 12 KB of cover information inside the cover image and the technique focused only on the payload[26]. Zang and Wang et al. introduced proficient steganographic inserting by taking advantage of the alteration course. The author focused on embedding more messages and also try to get security[27].

Khan et al. offered a cyclic image-based steganography method using randomization. In this paper, authors struggled to get consistency among the basic criteria and embedded 4 to 8 KB secret messages. Karim suggested a new image steganography technique in light of LSB using a secret key and achieving the security, payload but broke down the other criteria[28]. Muhammad and Sajjad proposed and magic LSB method using multi-level encryption and HSI image. It is a better technique in image steganography but has some limitations to different attacks such as scaling, noising, cropping, etc.[29].

Rustad et al. developed a novel image steganography for improving the perceptual transparency of the image. It is an inverted LSB method using an adaptive pattern. This research achieved a high-quality stego image but the payload limit is too low[30].

Cheng and Huang proposed a novel method of reversible encrypted image-based steganography using interpolation image and histogram shifting. In this study, the author used double scrambles operation on the pixel of the image by changing the positions of the pixels. The experimental results achieved a high embedding capacity and security[31].

Shwe Sin et al. presents an LSB and Huffman code-based steganography. The investigational outcomes presented that the algorithm got high embedding capacity and better security[32].

Tsai et al. presented another LSB-based image steganography algorithm in light of MSB prediction and the Huffman base method. The main objectives of the paper are high payload and secrecy and to increase the inserting pace of the mystery message inside the cover picture[33].

M. Shahu et al. presented a novel technique, LBP-based reversible information stowing away effectively accomplishes better HC, SI quality, and strength to different assaults. Be that as it may, besides these benefits, the proposed method can be improved concerning HC. Since the proposed work utilizes the LBP-based strategy, subsequently, it considers the surface and smooth pictures as unclear while embedding the EBs. Regardless, the pixel power of surface pictures makes it more sensible to introduce more EBs when appeared differently in relation to smooth pictures. Thusly, the proposed work can be contacted to achieve higher HC in surface pictures by brushing LBP with the PVDS-based technique[34].

Dhivya et al. proposed a proficient variable bit information embedding method founded on combined chaotic system (CCS) and integer wavelet transform (IWT). With the plan to boost the security level, CCS is created by joining chaotic maps of two 1D. The developed combined tent-logistic (CTL), combined sine-logistic (CSL) and combined sine-tent (CST) maps with improved chaotic behavior are utilized to generate the key sequences. From CTL, CSL and CS, these chaotic key sequences are then quantized to embed the secret bits in high-high, high-low and low–high sub-bands of the IWT transformed cover image, respectively. The fundamental benefit of this plan is that the number of bits to be implanted in every single sub-band coefficient is profoundly chaotic and very delicate to the underlying seeds[35].

Ramapriya et al. proposed a clinical picture steganography technique by taking advantage of Double Tree-Complex Wavelet Change based change and picture encryption system. Then a better SSOA enhancement calculation is locked in to distinguish smooth edge blocks. Subsequently, the determination of pixels for inserting is worked with. Installing the Restricted information into the cover picture is then done utilizing a twofold network XOR encoding. After the inserting system, the stego picture is created. Subsequently, the proposed strategy shows the best outcomes with high pay load limit, security, and picture quality than the current techniques. Testing was performed on PSNR, MSE, IF, and SSIM measurements to confirm the presentation of the proposed techniques[36].

Sahu et al. proposed a better method for two delicate watermarking plans to perform altering recognition and limitation in a picture. The proposed plans were outwardly disabled and the watermark pieces were created using the turbulent system-based determined map at both the source and getting end. Further, the chief contrives saw a constraint of ± 1 contrast between the host and watermarked pixels. As such, the idea of the watermarked picture was far superior to that of various plans. Finally, surprising results were achieved in regard to the adjusted area and limitation limit of the proposed plans. Later on, the proposed work can be connected with self-recovery of the modified pieces, and growing both spatial and change spaces to extra overhaul the force. There are different reported works presented in the literature based on image steganography, and each method has its related advantages and disadvantages depending on its embedding procedures and selection of the cover object[36]. Moreover, every examination work has attempted to cover the rudiments standards of picture steganography such are payload, strength, discernment, temper insurance, and computation. Some examination of existing techniques is introduced in Table 1.

Table 1 Critical analysis of various existing methods using criterion of image steganography.

In sum-up, Table 1 illustrated the analysis of the diverse existing methods using the basic criterion of steganography. It also expounded the techniques used, pros and cons, main focus, embedding procedure, and limitations of each method. Though, after a detailed analysis of the existing methods many points come to mind, but some point is very vital; one is selecting an appropriate cover object and the second is the reliability between the essential criterion of steganography. Because some tried for making a reliable method but get one or two parameters but broke down the other criteria and also the reliability between the criterion which is an essential part of steganography. However, to tackle these vital needs, the proposed algorithm is designed in a manner that shows which dimension of the image is better for which size of secret message and also to make a reliable method. However, the detailed whole process of the proposed algorithm is presented in the next section.

This section presents the whole process of the proposed algorithm, embedding and extraction process, Magic matrix, MLEA, Huffman encryption, and extraction process shown in Fig. 3. The mathematical notations which are used throughout the paper are presented in Table 2.

Graph: Figure 3The process of encrypted algorithm used in the proposed method.

Table 2 Mathematical notations used in the proposed algorithms.

The step-by-step procedures of embedding, the extraction process, MLEA, Magic Matrix (MgMtx), and Huffman code algorithms are presented given in Algorithms 2, 3, 4, and 5 respectively. The proposed algorithm used two stages in which first we encrypt the mystery message using Huffman code priority. For mapping the one secret word to one code word the optimal codes are Huffman code mapping. Since Huffman designates a parallel worth or code to each brightness of the cover picture. For applying the Huffman code, first, we convert the 2D image with size C × R to 1D array bits' stream with length of LHuff fewer than or equal to C × R (LHuff ≤ C × R). The given bit's stream is used for building the Huffman table or sequences HFCS because extracting the original message bits' stream is totally dependent on the same Huffman building table bits' stream. To lessen the size of the cover picture performs lossless pressure of the Huffman inserting is utilized. After the size of that image is reduced because Huffman encrypts shrinkages the image size. In the context of embedding, Huffman encrypted bit stream cannot reveal or show the message bits of the cover image or anything about it. Huffman is one type of verification because the secret message bit stream is only extracted using the same Huffman table or sequences which are used for embedding. If a little bit of error occurs the Huffman table or sequences can't extract and is unable to recover the secret message. The proposed method used MLEA, Magic Matrix, and Huffman code-based encryption to prove the efficiency, effectiveness, and dominancy of the proposed. Let's first elaborate on numerous important concepts which are used in the proposed algorithm.

The Multi-Level Encryption (MLE) is used for the proposed algorithm having different encryption operations to improve the security and give hardness to the attackers[49],[53]. It is used on secret data before embedding to the cover object or image for increasing security using some XOR operation to make the message unrecoverable for any assailants. While Magic matrix is used to make a matrix having no repeated numbers and the summation of that matrix either each row, each column, and both diagonals are remaining the same. It can be used for applying different operations on secret data or cover images due to its properties in terms of improving security and also giving the hard-hitting time to attackers shown in Figs. 4, 5, 6, 7 and 8 sequentially[47],[48],[54]. Let's try to elaborate with the help of the example given below.

Graph: Figure 4An example of the magic matrix (Steps: 1 & 2).

Graph: Figure 5An example of the magic matrix example (Step: 3).

Graph: Figure 6An example of the magic matrix example (Steps: 4a & 4b).

Graph: Figure 7An example of the magic matrix example (Step: 5).

Graph: Figure 8An example of the magic matrix example (Step: 6).

However, the Magic matrix is a special type of MATLAB function used for applying different shuffling, rotation, or reflecting properties. The above example in Figs. 4, 5, 6, 7 and 8 respectively shows us the detailed explanation of the matrix by taking the magic matrix 3 by 3 and the required magic sum is 15. Now perform different calculations on the matrix to make the matrix magic. Either trying from the center point, starting left–right corners, or even–odd numbers to make the matrix magic. So take the value in terms of 3 by 3 and make the summation diagonally, each row and column, and check the properties of the magic matric, if found correct then the matrix is said to be magic shown in Figs. 5, and 6. The above example from steps 1–6 elaborates on the step by steps method that how to making the matrix magic. You can see in the above Fig. 6, taking the square center to be 5 but the center involved 4 different sums, and if we take even numbers from the corners then each corner involved 3 different sums shown in step 4. And if we take the edges to be odd numbers then each sums involved 2 different sums. Now try to with some other numbers; suppose can we put 1 in 4 spots, or put 3 in 2 spots, so if we 2 multiply 4 then 8 possibilities occur. So there are 8 possibilities to put odd numbers on edges because we tried different ways but the result is not near to the magic matrix properties with odd numbers the different sums are low so we take the odd edges shown in step 5. Finally, the even numbers are placed at corners, and the odd numbers are placed on edges.

Graph

Graph

Graph

Graph

Graph

The section demonstrates the investigational results based on different perspectives, and some statistical assessment matrices to demonstrate the effectiveness and enactment of the planned work. For analysis, we used some standard images namely, Lena, Mandrill, Girl, etc. for the analysis of the method shown in Fig. 9. It is also critically analyzed in relation to safety and enactment analysis based on three distinct perspectives shown in Fig. 10.

Graph: Figure 9Datasets of standard images [70] , [71].

Graph: Figure 10Experimental evaluation perspectives for the proposed algorithm.

Before going to the performance and security analysis of the proposed algorithm, we first explain the Quality assessment metrics used for critical analysis of the proposed algorithm shown in Table 3[55]–[57].

Table 3 QAM's for the proposed algorithm[55]–[57].

The performance analysis using P1 is given as follows. Using P1 the planned work is evaluated on different images of distinct dimensions with 14 kb's secret message. Other hand, it is also analyzed based on different sizes of secret messages embedded into similar dimension images. So, the results demonstrate the connotation of the technique in view of PSNR values displayed in Table 4. While Table 5 shows the performance of the proposed method compared with relevant existing methods using PSNR values which 6.22% outer performed.

Table 4 Investigation of the proposed algorithm based on P 1.

Table 5 Average results of the proposed algorithm based on PSNR.

Security Analysis of the Proposed algorithm based on P2 is given. Table 6 elaborate the significance, resistance against to different attacks, quality, and performance of the propose method using different quality assessment parameters.

Table 6 Security analysis of the proposed algorithm based on Perspective 2.

In this part, we described some enactment of the proposed method. Figure 11a and d address the cover picture through the outcome tests. Figure 11b, and e are stego pictures that are completely impalpable. Figure 11c and f are decoded pictures that are vague to cover pictures. We preference the normal selection of cover pictures ''peppers'' and ''lake'' with the size of 256 × 256 as experimental tests.

Graph: Figure 11Experimental results of cover images (a, d), cipher images (b, e) and decrypted images (c, f).

To analyze whether the proposed encryption method can oppose differential assaults; two significant assessment factors for differential assaults investigation are used which is Unified Average Changing Intensity (UACI), and Number of Pixels Changing Rate (NPCR)[58],[59]. The value of UACI and NPCR of "Lake" and "Pepper" are presented in Tables 7 and 8 below. We can see that UACI is close the hypothetical value of 34.5742% and NPCR is near the hypothetical value of 99.7183%, and that implies that our plan can oppose difference assaults. In the meantime, our calculation is better than the literature[60],[61].

Table 7 UACI value of the proposed method.

Table 8 NPCR value of the proposed method.

Now to check the proposed algorithm on differential attack two attacks are; Noise Attack (NA), NA is used to check either the algorithm can resist against some noise attacks are not, because a good embedding algorithm should be able to resist against NA. We analyzed some standard images using noise attacks with the value of 0.01, 0.1 and 0.5 of salt and pepper noise. It can be seen in Fig. 12 by adding 0.1 salt and pepper noise are still detectible. Therefore, our algorithm has good toughness and can proficiently oppose commotion assaults.

Graph: Figure 12Noise attack (NA) analysis: (a) is decrypted image of (d, b) decrypted image of (e, c) decrypted image of (f, d) adding 0.01 salt and pepper noise, (e) adding 0.01 salt and pepper noise, (f) adding 0.01 salt and pepper noise.

While Cropping Attacks (CA), An ideal cryptosystem ought to be against information CA by transmission and capacity[62],[63]. To assess its power in opposing trimming assaults, leaves behind 64 × 64, 64 × 128, 128 × 128, and 128 × 256 are obliterated from the embedded image ''Lake '' as displayed in Fig. 13e–h. The extracted image is displayed in Fig. 13a–d, and they can still be perceived. It demonstrates that our algorithm can oppose information editing assaults.

Graph: Figure 13Cropping Attack (CA) analysis (a) decrypted image of (e, b) decrypted image of (f, c) decrypted image of (g, d) decrypted image of (h, e) 1/16 CA, (f) 1/8 CA, (g) 1/4 CA, (h) 1/2 CA.

Histogram Analysis of the proposed algorithm based on P3 is given. Histogram Examination shows the real contrast between both stego and cover images. Due to its properties, a little difference between images can be founded, because it shows the extracted occurrences of the pixels of the image. Figure 14 shows the analysis of the proposed algorithm using three standard images namely, Lena, Mandrill, and House as both stego and cover image histograms.

Graph: Figure 14Histogram analysis of the proposed algorithm based on three standard cover and stego images.

However, the empirical results of the proposed algorithm grounded on diverse viewpoints using assessment metrics prove the improvement, efficiency, and effectiveness of the method. Our method improves the payload, high-level security, temper protection, and better visual quality of the image. Suppose attackers or any naked eye suspects the technique used in the proposed algorithm which is LSB and some improved technique of steganalysis. Then attacker can't extract the actual contents of the secret message because the proposed method used Huffman code[64],[65], MLE[66], Magic matrix[67],[68], and HSI (achromatic components of an image)[69]. If an attacker extracts the message up to some limits then the contents are useless, because, for extraction of the full message text, attackers need to use Huffman code, MLE, etc. for getting the actual message.

In this study we proposed a novel method using Huffman code, HSI color model, MLEA, Magic matrix, and LSB substitution. Embedding the secret message, the I-plane of the HSI variety model is utilized as the cover image rather than the RGB model, for expanding the safety and reducing extra computational upstairs or processing time. The empirical outcomes of the proposed technique secure a normal of PSNR 79.29 dB over 165 standard images and also proves the control, and efficiency of the proposed method compared with some stated correlated works. The uses of the way of embedding the secret message in cover object is to makes this algorithm dreadful and abstruse and also unclear and foolish the steganalysis process. Our proposed method also used Huffman code which makes it more robust than existing methods. However, we infer that it is capable of shaping the stego picture practically identical and also has the fitness to give adeptness, and effectiveness and justify the encouraging demands of the current system and user to generate better quality stego images. So our method is easy to program, simple, and a better combination in terms of transparency and robustness. The results from different perspectives show the achievability and outperforming of our method to others. The main demerit of the proposed method is the amount of embedding the secret message not more than 20 KBs. Because for reliability, we analyzed the algorithm from different perspectives to achieve the basic criterion of steganography up to some acceptable limits. But still needs some reasonable improvements by functioning on magic matrix and MLEA extension to make the technique more dependable and also implementation of the method into the transform domain[70],[71]. We are also now involved in an outcome further such ways as unsupervised learning (ML), related concepts of Deep Learning to grab some boundaries, and statistical and image processing assaults to produce excellent and dependable free stego images.

This research is supported, in parts, by the University of Peshawar and, in parts, by the Abdul Wali Khan University Mardan (AWKUM), Pakistan.

Shahid Rahman:- Research, Methodology, Writing - Original Draft; Jamal Uddin:- Conceptualization, Methodology, Software, Writing - Review & Editing; Hameed Hussain:- Conceptualization, Visualization, Validation, Investigation; Aftab Ahmed:- Visualization, Validation, Proofreading; Ayaz Ali Khan:- Writing - Review & Editing, Revisions; Muhammad Zakarya:- Visualization, Data Curation, Proofreading; Afzal Rahman:- Visualization, Writing - Review & Editing; Muhammad Haleem:- Writing - Revised Draft, Data Curation;

The datasets produced and/or analyzed during the current study are openly accessible in the Kaggle repository, and SIPI, and can be gotten to at [https://www.kaggle.com/datasets/mnavaidd/image-segmentation-dataset]. Further, different pictures utilized inside the exploratory work are freely accessible on the web. All the codes used for this method will be provided for research purposes if requested by researchers.

The authors declare no competing interests.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.