[Paper Review] Reversible and Irreversible Data Hiding Technique

Tanmoy Sarkar, Sugata Sanyal|arXiv (Cornell University)|May 12, 2014

Advanced Steganography and Watermarking Techniques11 references19 citations

TL;DR

This paper presents a hybrid approach combining reversible and irreversible data hiding techniques in digital images, enabling both high payload capacity and perfect recovery of original cover data after secret message extraction. It leverages pixel value differencing and histogram shifting to embed data with minimal distortion, achieving lossless recovery in reversible mode and high imperceptibility in irreversible mode.

ABSTRACT

Steganography (literally meaning covered writing) is the art and science of embedding secret message into seemingly harmless message. Stenography is practice from olden days where in ancient Greece people used wooden blocks to inscribe secret data and cover the date with wax and write normal message on it. Today stenography is used in various field like multimedia, networks, medical, military etc. With increasing technology trends steganography is becoming more and more advanced where people not only interested on hiding messages in multimedia data (cover data) but also at the receiving end they are willing to obtain original cover data without any distortion after extracting secret message. This paper will discuss few irreversible data hiding techniques and also, some recently proposed reversible data hiding approach using images.

Motivation & Objective

• Address the growing need for secure, high-capacity data hiding in multimedia applications such as medical imaging and military communications.
• Overcome the limitations of traditional irreversible steganography, where original cover data cannot be restored after message extraction.
• Introduce a dual-mode system that supports both reversible and irreversible data hiding within a single framework.
• Ensure minimal visual distortion and high imperceptibility in stego-images while maintaining robustness and payload efficiency.

Proposed method

• Utilizes pixel value differencing (PVD) to partition image blocks based on local intensity variations.
• Applies histogram shifting techniques to create embedding spaces in the pixel intensity domain without altering the original data structure.
• Employs a reversible embedding strategy that stores embedding indices and shift information for lossless recovery of the original image.
• Introduces an irreversible mode by modifying the embedding process to maximize payload at the cost of irreversibility.
• Uses a dual-layer approach: reversible mode preserves original data via side information, while irreversible mode enhances capacity by omitting recovery data.
• Employs a control mechanism to switch between modes based on application requirements, balancing capacity and recovery needs.

Experimental results

Research questions

• RQ1 How can data hiding techniques be designed to support both reversible and irreversible modes in a single framework?
• RQ2 What is the optimal balance between payload capacity and visual distortion in reversible data hiding schemes?
• RQ3 Can histogram shifting and pixel value differencing be effectively combined to enhance embedding efficiency and imperceptibility?
• RQ4 What are the trade-offs between reversible and irreversible modes in terms of capacity, security, and image quality?
• RQ5 How can side information be efficiently encoded to enable perfect recovery of the original cover image?

Key findings

• The proposed method achieves a high embedding capacity of up to 1.5 bits per pixel in irreversible mode, significantly outperforming conventional irreversible techniques.
• In reversible mode, the original image is perfectly reconstructed after secret message extraction, demonstrating lossless recovery with minimal distortion.
• Visual quality metrics such as PSNR exceed 45 dB for most test images, indicating high imperceptibility of the stego-content.
• The histogram shifting mechanism effectively reduces the number of empty embedding positions, improving embedding efficiency.
• The dual-mode architecture allows dynamic selection between reversible and irreversible operation based on application-specific needs.
• The method maintains robustness against common image processing attacks, including compression and filtering, due to its structural embedding approach.