Reversible data hiding in encrypted images with multi-prediction and adaptive huffman encoding.

Area of Science:

• Computer Science
• Information Security
• Multimedia Technology

Background:

• Massive data generation and sharing raise concerns about inappropriate access and abuse.
• Reversible data hiding in encrypted images (RDHEI) offers privacy preservation by embedding data in an encrypted domain.
• Limited data embedding capacity is a major challenge for current RDHEI methods.

Purpose of the Study:

• To propose a novel RDHEI scheme that enhances data embedding capacity.
• To improve the performance and applicability of RDHEI techniques.
• To maintain the security and reversibility of the data hiding process.

Main Methods:

• Developed a multi-prediction strategy (MED+GAP predictor) for generating label map data before image encryption.
• Implemented adaptive Huffman coding to compress generated labels, reducing auxiliary information length.
• Evaluated the scheme on BOSSBase, BOWS-2, and UCID datasets.

Main Results:

• The proposed RDHEI scheme achieved average improvements of 0.052 bpp, 0.023 bpp, and 0.047 bpp over state-of-the-art methods on the tested datasets.
• The method effectively increased the overall data embedding capacity.
• Security and reversibility were successfully maintained.

Conclusions:

• The novel RDHEI scheme based on multi-prediction and adaptive Huffman coding offers a significant improvement in data embedding capacity.
• This advancement addresses a key limitation in RDHEI, paving the way for broader applications.
• The method provides a secure and reversible solution for protecting data in encrypted images.