Abstract
A novel data hiding scheme is proposed based on encoding and substitution. Both the cover media and the secure information are encoded to achieve the substitution. Simulating the genetic codon table in molecular biology, a random grouping algorithm is proposed to generate the pseudo genetic codon table, which regulates the substitution. The cover media is encoded into a DNA codon like sequence. Then, the codons of the cover media in the form of DNA codon sequence are substituted with other codons in the same group according to the pseudo genetic codon table. The secure information in the arbitrary N-nary form decides which codon is selected during every substitution. Due to the random construction of pseudo genetic codon table, encryption is incorporated in the data hiding process. Analysis and experiments have tested the validity and efficiency of the scheme.
Introduction
Data hiding, or steganography, is the practice of concealing secret messages within digital media such as images, audio, or video. The primary challenge in this field is the "steganographic trilemma," which requires balancing high data capacity, high visual fidelity (quality), and robust security. Traditional methods, such as Least Significant Bit (LSB) substitution, often fail to meet all three criteria simultaneously and are frequently vulnerable to statistical analysis.
Furthermore, many existing DNA-based steganography techniques rely on fixed biological tables or require the original cover image for message extraction (non-blind extraction). This paper introduces a bio-inspired scheme that utilizes a randomized "pseudo genetic codon table" and Module-N operations to achieve high capacity, superior visual quality, and blind extraction, where the receiver only needs the secret keys to recover the message.
Methodology
The proposed approach simulates the biological process of genetic translation to hide data within digital images through a series of sophisticated encoding steps.
DNA Encoding and Codon Grouping
First, the cover image is converted into a sequence of nucleotides (A, C, G, T) based on its binary pixel values. These nucleotides are then grouped into triplets called "codons." In nature, 64 possible codons map to only 20 amino acids, creating redundancy where multiple codons (synonyms) represent the same amino acid. The researchers replicate this by randomly partitioning the 64 digital codons into $N$ groups. This randomized grouping, known as the pseudo genetic codon table, acts as a critical secret key.
N-nary Transformation and Substitution
The secret message is converted from binary into an N-nary numbering system, where the base $N$ corresponds to the number of codons available in each group of the pseudo-table. To hide the data, the algorithm identifies a codon in the cover image, finds its corresponding group in the pseudo-table, and replaces it with another codon from that same group. The specific replacement is determined by the secret data value using a Module-N operation. This ensures that the modification is subtle and mathematically precise.
Experimental Results
The scheme was evaluated using standard benchmarks, including grayscale images from the USC-SIPI database. Experiments conducted on 512×512 images demonstrated that the DNA-inspired method significantly outperforms standard LSB techniques in both security and fidelity.
At an embedding rate of 1.5 bits per pixel (bpp), the proposed scheme achieved a Peak Signal-to-Noise Ratio (PSNR) of approximately 46-47 dB, compared to the 45 dB typically seen with LSB methods. Higher PSNR values indicate that the hidden data is less detectable to the human eye. Additionally, the scheme reached embedding rates as high as 2.0310 bpp while maintaining a high PSNR above 43 dB, with an embedding efficiency of 3.6134 bits hidden per unit of image change.
Conclusion
By leveraging the redundancy of the genetic code and the randomness of a pseudo-codon table, this scheme provides a secure and efficient method for data hiding. The integration of encryption into the embedding process and the ability for blind extraction represent significant advancements over traditional steganographic methods. Future work will explore the robustness of this scheme against lossy compression attacks, such as JPEG processing, to ensure reliability in real-world network environments.
Summary updated