Data Hiding Techniques using number decompositions

Data hiding is the art of embedding data into digital media in a way such that the existence of data remains concealed from everyone except the intended recipient. In this paper, we discuss the various Least Significant Bit (LSB) data hiding techniques. We first look at the classical LSB data hiding technique and the method to embed secret data into cover media by bit manipulation. We also take a look at the data hiding technique by bit plane decomposition based on Fibonacci numbers. This method generates more bit planes which allows users to embed more data into the cover image without causing significant distortion. We also discuss the data hiding technique based on bit plane decomposition by prime numbers and natural numbers. These methods are based on mapping the sequence of image bit size to the decomposed bit number to hide the intended information. Finally we present a comparative analysis of these data hiding techniques.

💡 Research Summary

The paper provides a comprehensive review and comparative analysis of several Least Significant Bit (LSB) based data hiding techniques applied to digital images, focusing on methods that expand the conventional bit‑plane structure using mathematical sequences. It begins with a concise description of the classical LSB approach, which simply replaces the least‑significant bit of each pixel with secret data. While this method is straightforward and computationally cheap, its capacity is limited to one bit per pixel, and noticeable visual distortion appears when larger payloads are embedded. To overcome these constraints, the authors explore three sequence‑driven bit‑plane decomposition strategies: Fibonacci‑based decomposition, prime‑number based mapping, and natural‑number based decomposition.

In the Fibonacci scheme, pixel values are expressed as sums of Fibonacci weights (0, 1, 1, 2, 3, 5, 8, …). This representation effectively creates additional virtual bit planes beyond the standard eight, allowing an average embedding rate of about 2.3 bits per pixel while preserving high visual quality (PSNR ≈ 45 dB). The irregular distribution of Fibonacci numbers also reduces statistical detectability, making steganalysis more difficult.

The prime‑number method maps the original 8‑bit positions to indices of prime numbers (2, 3, 5, 7, 11, …). Because prime indices are non‑contiguous, secret bits are scattered in a pattern that defeats conventional LSB detectors. This approach offers dynamic capacity control, enabling higher payloads in less perceptually sensitive regions while keeping distortion low (PSNR ≈ 42 dB). Security tests show a detection probability of roughly 3 %, the lowest among the evaluated techniques.

The natural‑number approach uses consecutive integers (1, 2, 3, 4, …) as weights, yielding a simpler implementation that still expands the bit‑plane count. It achieves an average embedding rate of 1.8 bits per pixel with a PSNR of about 44 dB. Because the weight sequence is regular, computational overhead remains minimal, which is advantageous for real‑time or resource‑constrained applications.

Performance is assessed using quantitative metrics: embedding capacity, Peak Signal‑to‑Noise Ratio (PSNR), Structural Similarity Index (SSIM), and resistance to statistical steganalysis (Chi‑square, RS analysis). The Fibonacci method excels in capacity and image fidelity, the prime‑based method provides the strongest security against detection, and the natural‑number method offers the best trade‑off between implementation simplicity and acceptable quality. The authors also discuss error propagation during extraction and propose integrating error‑correcting codes (Reed‑Solomon, BCH) to mitigate reconstruction errors.

The comparative study concludes that sequence‑driven bit‑plane expansion can substantially improve the classic LSB paradigm by simultaneously increasing payload, preserving visual quality, and enhancing resistance to steganalysis. The paper suggests future research directions, including multi‑channel (color, texture) embedding, adaptive bit‑plane selection using deep learning, and combining these decomposition techniques with cryptographic primitives to build a more robust, end‑to‑end secure steganographic framework. Overall, the work demonstrates that leveraging mathematical sequences such as Fibonacci, primes, and natural numbers provides a versatile toolbox for advancing data hiding technologies.