Flexible visually secure image encryption with meta-learning compression and chaotic systems.

Area of Science:

• Computer Science
• Information Security
• Artificial Intelligence

Background:

• Growing demand for secure digital image encryption across various applications.
• Limitations of existing methods include insufficient security and poor decrypted image quality.
• Need for advanced techniques integrating compression and encryption for visual data.

Purpose of the Study:

• To propose a flexible and secure image encryption scheme.
• To enhance decrypted image quality and encryption security.
• To leverage meta-learning, chaotic systems, and deep learning for image security.

Main Methods:

• Developed a meta-learning compression reconstruction network with dynamic auxiliary input for high-quality image compression.
• Constructed a novel IS-DP chaotic system by combining 2D-IS chaotic system with a deep learning network for image encryption.
• Implemented a lossless LSB-2^k correction embedding method for embedding the secret image into a carrier image.

Main Results:

• Achieved high-quality compression and visually secure encryption of digital images.
• Demonstrated the effectiveness of the proposed IS-DP chaotic system and meta-learning approach.
• Validated the feasibility of deep learning methods in integrated encryption and compression tasks.

Conclusions:

• The proposed scheme offers a flexible solution for visually secure image encryption.
• Meta-learning provides adaptability, allowing users to balance performance and quality.
• The integration of deep learning and chaotic systems shows significant potential for advanced image security applications.