Reversible data hiding with dual pixel-value-ordering and minimum prediction error expansion.

Area of Science:

• Digital Image Processing
• Information Security
• Data Hiding Techniques

Background:

• Reversible Data Hiding (RDH) is crucial for embedding information without degrading image quality.
• Pixel Value Ordering (PVO) is a known technique for high-fidelity RDH.
• Existing PVO-based methods have limitations in rate-distortion performance.

Purpose of the Study:

• To introduce a novel dual Pixel Value Ordering (dPVO) method for Prediction Error Expansion (PEE).
• To develop an improved RDH scheme with enhanced rate-distortion performance.
• To achieve better embedded image quality at higher embedding rates.

Main Methods:

• A two-phase embedding process: forward and backward.
• Forward phase: Classic PVO with PEE on 1x3 image blocks.
• Backward phase: dPVO applied to minimum and maximum pixel sets to refine predictions.

Main Results:

• The proposed dPVO with PEE partially restores predicted pixel values.
• Achieved improved embedded image quality and higher embedding rates.
• Demonstrated superior rate-distortion performance compared to state-of-the-art PVO-based RDH schemes.

Conclusions:

• The novel dPVO scheme offers significant improvements in RDH.
• This method provides a better balance between embedding capacity and image fidelity.
• The dPVO technique represents a promising advancement in reversible data hiding.