Source-linked AI summary
A novel magic LSB substitution method (M-LSB-SM) using multi-level encryption and achromatic component of an image
Khan Muhammad, Muhammad Sajjad, Irfan Mehmood, Seungmin Rho, Sung Wook Baik
TL;DR
Image steganography seeks to conceal secret data while limiting statistical detectability, but existing methods face security and imperceptibility limitations. This paper proposes M-LSB-SM, combining HSI I-plane embedding, four-block MLEA encryption, rotated sub-images, and Magic LSB substitution. Across evaluations, the method reports strong image quality and security, including an average PSNR of 47.93 dB over 150 images, while identifying payload efficiency and transform-domain robustness as future improvements.
Problem
Existing steganographic methods have limitations in security and imperceptibility, while direct unencrypted embedding can make secret extraction easier if the algorithm is compromised.
Method
M-LSB-SM transposes and converts RGB images to HSI, encrypts the message with MLEA into four blocks, rotates four I-plane sub-images, and embeds corresponding blocks using Magic LSB substitution.
Results
The proposed scheme achieves an average PSNR of 47.93 dB over 150 images and is reported to outperform the compared schemes in image-quality measures.
Takeaways & Limitations
The authors conclude that M-LSB-SM offers a combination of imperceptibility and security suitable for steganographic applications.
Takeaways & Limitations
Future work targets improved payload efficiency, stronger MLEA, and transform-domain implementation for resilience against image-processing and statistical attacks.
Abstract
from arXiv · showhide
Image Steganography is a thriving research area of information security where secret data is embedded in images to hide its existence while getting the minimum possible statistical detectability. This paper proposes a novel magic least significant bit substitution method (M-LSB-SM) for RGB images. The proposed method is based on the achromatic component (I-plane) of the hue-saturation-intensity (HSI) color model and multi-level encryption (MLE) in the spatial domain. The input image is transposed and converted into an HSI color space. The I-plane is divided into four sub-images of equal size, rotating each sub-image with a different angle using a secret key. The secret information is divided into four blocks, which are then encrypted using an MLE algorithm (MLEA). Each sub-block of the message is embedded into one of the rotated sub-images based on a specific pattern using magic LSB substitution. Experimental results validate that the proposed method not only enhances the visual quality of stego images but also provides good imperceptibility and multiple security levels as compared to several existing prominent methods.
1. Introduction
The introduction reviews limitations of existing steganographic methods and presents M-LSB-SM as a more secure, imperceptible approach using HSI, encryption, and structured embedding.
- Spatial-domain methods offer high payload but remain vulnerable to image-processing manipulations and statistical attacks.
- Classical LSB substitution increases payload by replacing more least significant bits, but introduces noticeable carrier-image changes.
- LSB-MR embeds two secret bits through relationships between paired pixels, reducing modification from 0.5 to 0.375 bpp at the same capacity.
- Edge-based methods exploit larger differences between neighboring pixels because edge pixels can carry more secret bits than smooth pixels.
- M-LSB-SM combines HSI achromatic-plane embedding, MLEA encryption, and four-block Magic LSB embedding to increase security and complicate extraction.
2. Literature Review
The literature review explains basic LSB embedding, its payload–distortion trade-off, and increasingly structured alternatives that address quality, security, or payload limitations.
- 2.1 Basic Idea of LSB methods: Classical LSB substitution replaces host-image least significant bits with secret-data bits to produce the stego image.
- 2.1 Basic Idea of LSB methods: With k=1, only half of the illustrated pixels change and asymmetry artifacts are nearly imperceptible to the human visual system.
- 2.1 Basic Idea of LSB methods: Increasing the number of replaced LSBs raises payload but causes obvious distortion in the stego image.
- 2.2 LSB-M and LSB-MR: LSB-M randomly adds or subtracts 1 when a message bit mismatches the pixel LSB, while LSB-MR embeds paired bits through relationships between two pixels.
- 2.3 Pixel Indicator based Methods: Pixel-indicator methods use an indicator channel to select data channels, improving payload and security but potentially producing lower payload or noticeable distortion.
- 2.4 Edges based Data Hiding Methods: Existing methods often embed directly without encryption, making extraction easier when the algorithm is compromised; M-LSB-SM addresses this with four-block MLEA encryption and multiple barriers.
3. The Proposed Scheme
The proposed M-LSB-SM scheme combines HSI I-plane processing, key-based image subdivision, multi-level encryption, and magic LSB embedding. It encrypts and disperses four message blocks across rotated sub-images before reconstructing the stego image.
- Scheme overview: M-LSB-SM transposes the cover image, converts it from RGB to HSI, and uses the achromatic I-plane for embedding.The paper states that modifying the I-plane avoids changing the other HSI planes and is relatively cost effective.
- Multi-level encryption: MLEA encrypts the secret data into four distinct cipher blocks using the secret key before embedding.The encryption procedure divides message bits into B1, B2, B3, and B4, then applies multiple bitwise and permutation operations.
- Embedding pipeline: The I-plane is divided into four equal sub-images, each rotated by a key-dependent angle and paired with one encrypted message block.The corresponding sub-images and message blocks are processed through the ImageSubDivision and MagicLSB stages.
- Magic LSB: Magic LSB uses a magic matrix to determine where successive secret bits are stored within each sub-image.In the 3×3 example, the matrix assigns the eight message bits to a dispersed sequence of pixel locations rather than embedding them sequentially.
- Extraction: The extraction process reverses transposition, HSI conversion, sub-image decomposition, key-based rotations, message extraction, and decryption.The extracted blocks are decrypted to recover the actual secret message.
4. Experimental Results and Discussion
The study evaluates the proposed scheme across three perspectives using 150 images, multiple IQAMs, and comparisons with seven existing methods. Results indicate strong imperceptibility, visual quality, and security, with the proposed method generally outperforming competing approaches.
- Experimental setup: The evaluation uses 150 images from USC-SIPI-ID and COREL datasets, covering diverse image content and standardized resolutions.The datasets include edgy and smooth color images, with dimensions adjusted for consistent analysis.
- Experimental setup: Three perspectives vary payload size, image content, and image resolution while assessing the methods with multiple image-quality metrics.Perspective 1 embeds 8KB in 256×256 images; perspective 2 varies cipher size; perspective 3 varies resolutions while keeping the cipher at 8KB.
- Evaluation metrics: The comparison reports PSNR, SSIM, NCC, and MAE for the proposed method and seven existing schemes.The evaluated baselines include classical LSB, SCC, PIT, FMM, Karim's method, CST, and SHSI.
- Perspective 1: Across perspective 1, the proposed scheme achieves the highest values of the reported IQAMs, including averages computed over 150 images.The comparison embeds 8KB of text into diverse 256×256 images.
- Perspective 2: The proposed approach produces higher average PSNR and NCC than the mentioned alternatives across images with different visual characteristics.The evaluation includes smooth images such as Lena and edgy images such as Baboon.
- Perspective 3: For different resolutions, the proposed scheme provides promising PSNR results while embedding an 8KB text file.Perspective 3 tests 128×128, 256×256, 512×512, and 1024×1024 images.
- Qualitative and security analysis: Qualitative analysis reports high-quality stego images that are difficult to detect visually compared with other methods.The paper attributes the proposed method's security and imperceptibility to encryption, block division, HSI conversion, rotations, and Magic LSB embedding.
- Performance analysis: The methods generally use a 1 bpp payload, while PIT exceeds this payload and FMM depends on window size, often remaining below 1 bpp.The payload comparison distinguishes capacity from imperceptibility and visual quality.
5. Conclusion and Future Directions
M-LSB-SM combines the HSI achromatic component with multi-level encryption to improve imperceptibility and security in color-image steganography. The authors report high average visual quality and identify payload efficiency, stronger encryption, and transform-domain implementation as future directions.
- Conclusion: M-LSB-SM uses the HSI achromatic component and multi-level encryption to provide improved imperceptibility and security for color images.The method divides secret information into four sub-blocks and processes them through MLEA before embedding.
- Conclusion: 47.93 dB average PSNR was achieved across 150 images, supporting the reported visual quality of the proposed scheme.
- Conclusion: The authors describe the algorithm as simple, easy to implement, and feasible for steganographic applications because it combines imperceptibility and security.
- Future directions: Future work targets improved payload efficiency, a more powerful MLEA, and transform-domain implementation for resilience against image-processing and statistical attacks.