A source-channel coding approach to digital image protection and self-recovery

Area of Science:

• Digital Image Forensics
• Information Security
• Error Correction Coding

Background:

• Watermarking algorithms are crucial for image forensics, particularly in detecting and recovering image tampering.
• Existing methods using check and reference bits struggle with recovering lost reference bit information after tampering.
• Image tampering poses a significant challenge to data integrity and authenticity.

Purpose of the Study:

• To develop a watermarking algorithm that effectively detects tampered image areas and recovers lost information.
• To model image tampering as an erasure error problem for improved data recovery.
• To enhance the quality of both watermarked and recovered images.

Main Methods:

• Designing a watermarking algorithm that allocates bit-budget to source encoder output, channel code parity bits, and check bits.
• Utilizing channel coding to protect reference bits against tampering, treating it as an erasure error.
• Employing check bits for tampering detection and erasure location identification.

Main Results:

• The proposed scheme significantly outperforms existing techniques in watermarked and recovered image quality.
• Improved image recovery quality is achieved through the consistent performance of source and channel codes.
• Watermarked image quality is enhanced by optimizing bit-budget allocation.

Conclusions:

• The proposed watermarking method effectively addresses image tampering by combining detection and recovery.
• Modeling tampering as an erasure error and applying channel coding offers a robust solution for data recovery.
• The technique provides superior image quality for both watermarked and recovered images, advancing image forensics capabilities.