Reversible Data Hiding via Integer Wavelet Coefficient Sign Modification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data hiding techniques often result in non-reversible distortion of the cover media, which is unacceptable for sensitive applications like legal and medical imaging, as they modify the original signal irreversibly during the data embedding process.
Innovation Solution
A reversible data hiding method based on integer wavelet spread spectrum and histogram modification, where data is embedded in the high-frequency sub-bands of the integer wavelet transform coefficients, and pseudo bits are used to flag coefficients, enhancing data hiding efficiency and preventing overflow/underflow by narrowing the histogram.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If data is embedded in the cover media using conventional data hiding techniques, then data embedding capacity is improved, but the original cover media cannot be completely restored (reversibility deteriorates)
Solution Approach 1:
The cover media is transformed into frequency domain segments (wavelet coefficients) where data can be embedded in specific high-frequency sub-bands without affecting the low-frequency components that carry the main image information. This segmentation allows independent modification of specific coefficient groups while preserving the overall structure for reversible reconstruction.
Solution Approach 2:
The invention changes the embedding domain from spatial domain to frequency domain (wavelet transform domain), and modifies specific parameters (high-frequency coefficients) rather than the entire image. By embedding data in least significant bits of selected coefficients and using histogram modification to prevent overflow/underflow, the system achieves both high embedding capacity and complete reversibility.
2Quantity of substance
If data is embedded in the cover media to increase embedding capacity, then more data can be hidden, but distortion of the cover media increases (visual quality deteriorates)
Solution Approach 1:
The invention applies different treatment to different frequency sub-bands: high-frequency coefficients are selected for data embedding while low-frequency coefficients are preserved unchanged. Within the high-frequency band, only specific coefficients are modified based on histogram analysis, ensuring that modifications are localized to regions where they cause minimal visual distortion.
Solution Approach 2:
The invention embeds data in only a portion of the available coefficients (those selected through histogram analysis) rather than modifying all coefficients. This partial action approach achieves sufficient embedding capacity while minimizing the total amount of modification and resulting distortion.
3Reliability
If histogram modification is applied to prevent overflow and underflow, then reversibility is maintained, but the complexity of the data hiding process increases
Solution Approach 1:
The histogram modification is performed as a preliminary step before data embedding. By pre-adjusting the histogram to prevent overflow and underflow conditions, the system ensures that subsequent embedding operations can proceed without compromising reversibility, simplifying the overall process flow.
Solution Approach 2:
The histogram modification acts as an intermediary process that mediates between the embedding operation and the reconstruction process. It creates a buffer zone in the coefficient distribution that prevents extreme values, thereby enabling lossless reconstruction without requiring complex compensation mechanisms.
Data Source
AI summary
A system and method are disclosed which may include subjecting an original, pixel domain image to an Integer Wavelet Transform (IWT) to obtain a matrix of IWT coefficients; selecting a plurality of the IWT coefficients for incorporation of information therein; and setting signs for the plurality of selected IWT coefficients according to bit values of a plurality of respective data bits. The system and method can also include subjecting a marked pixel domain image to an Integer Wavelet Transform to obtain a matrix of wavelet coefficients; selecting a plurality of the coefficients from the matrix that contain embedded information; and for each selected coefficient, extracting the data bit embedded in the coefficient, a bit value of the extracted data bit determined based on a sign of the coefficient.