Influential Metrics Estimation and Dynamic Frequency Selection Based on Two-Dimensional Mapping for JPEG-Reversible...

Area of Science:

• Digital Image Processing
• Information Security
• Steganography

Background:

• Existing JPEG Reversible Data Hiding (RDH) methods often fail to adequately manage file size increases alongside visual distortion.
• A need exists for RDH techniques that consider multiple critical metrics during data embedding.

Purpose of the Study:

• To develop a new JPEG RDH scheme that concurrently addresses visual distortion and file size increment (FSI).
• To introduce a dynamic frequency selection strategy for improved embedding efficiency and image recoverability.

Main Methods:

• A two-dimensional (2D) mapping strategy was employed to analyze non-zero alternating current (AC) coefficient pairs (NZACPs) and their run lengths.
• The method incorporates a pre-processing phase for blocks and a dynamic frequency selection strategy.
• Data embedding is performed using 2D histogram shifting (HS) after selecting optimal block groups based on computed metrics.

Main Results:

• The proposed JPEG RDH scheme demonstrated superior performance compared to existing methods.
• Consistent improvements were observed in both Peak Signal-to-Noise Ratio (PSNR) and optimized FSI.
• The method effectively balances data hiding capacity with image integrity and file size management.

Conclusions:

• The developed JPEG RDH scheme offers an effective solution for embedding data in JPEG images while minimizing negative impacts on file size and visual quality.
• This approach provides a more comprehensive optimization strategy for reversible data hiding in JPEG compression.
• The dynamic frequency selection and 2D mapping contribute to the method's overall efficiency and effectiveness.