Advanced EPISM approach for holographic stereogram generation utilizing neural radiance fields.

Area of Science:

• Computer Vision
• Holography
• Computer Graphics

Background:

• Traditional holographic stereogram printing faces challenges with inefficient light field acquisition and computational complexity.
• Existing methods often struggle with precise scene reconstruction and high-fidelity hologram generation.

Purpose of the Study:

• To develop an integrated approach for synthetic holographic stereogram printing using neural radiance fields (NeRF) and effective perspective images segmentation and mosaicking (EPISM).
• To address limitations in computational efficiency, light field acquisition, and reconstruction accuracy in current holographic technologies.

Main Methods:

• Trained a NeRF model with multi-resolution hash encoding using sparse perspectives for implicit scene representation.
• Utilized the EPISM method to calculate camera pose parameters and render parallax images via the trained neural network.
• Encoded rendered images using EPISM to generate synthetic effective perspective images for direct exposure printing.

Main Results:

• Demonstrated effective integration of EPISM and NeRF, overcoming challenges of traditional holographic printing.
• Achieved rapid computation, precise 3D scene reconstruction, and high-quality hologram printing.
• Provided an end-to-end solution for acquiring and printing high-fidelity synthetic holographic stereograms.

Conclusions:

• The combined EPISM and NeRF approach offers a significant advancement in synthetic holographic stereogram generation.
• This method provides a computationally efficient and accurate solution for creating realistic holographic displays from real-world scenes.
• The study paves the way for improved holographic printing technologies with enhanced fidelity and speed.