Light-field 3D encryption based on the monocular depth rendering.

Area of Science:

• Optics and Photonics
• Computer Vision
• Cryptography

Background:

• Traditional light-field 3D encryption faces limitations in real-time performance due to slow acquisition speeds and large data requirements for elemental image arrays (EIA).
• Existing encryption methods for 3D data often struggle with balancing security, efficiency, and reconstruction quality.

Purpose of the Study:

• To develop an advanced light-field 3D encryption scheme that overcomes the speed and data acquisition limitations of current integral imaging-based methods.
• To enhance the security and efficiency of 3D encryption through novel rendering and cryptographic techniques.

Main Methods:

• A novel light-field 3D encryption scheme utilizing monocular depth rendering from a single RGB image.
• Employing a convolution residuals network (CRN) to generate depth maps and simplify EIA acquisition via image mapping.
• Implementing reversible state loop cellular automata (RSL-CA) for encrypting single RGB images, updating traditional 3D encryption protocols.

Main Results:

• Successful generation of depth maps from single RGB images, simplifying the EIA pickup process.
• Demonstrated high security and efficiency in the encryption algorithm using RSL-CA.
• Experimentally verified clear, colorful optical 3D reconstruction with a notable parallax effect.

Conclusions:

• The proposed monocular depth rendering and RSL-CA based light-field 3D encryption scheme significantly improves real-time performance and security.
• This innovative approach offers a new research direction for developing advanced light-field 3D encryption technologies.
• The method provides a practical solution for secure and efficient 3D data representation and reconstruction.