Viewing resolution and depth-of-field enhancement for a digital 3D display based on neural network-enabled multilayer...

Area of Science:

• Optoelectronics
• Computer Vision
• Artificial Intelligence

Background:

• Traditional digital 3D displays face limitations in resolution and depth of field (DoF).
• These drawbacks stem from issues with transforming planar pixels into view perspectives and lens diffraction effects.
• These limitations hinder the widespread commercial adoption of 3D display technology.

Purpose of the Study:

• To enhance the viewing resolution and DoF of digital 3D displays.
• To propose a neural network-enabled method for optimizing elemental image arrays (EIAs).
• To overcome the limitations of current 3D display technologies.

Main Methods:

• A neural network-enabled multilayer view perspective fitting approach was developed.
• This method fits reconstructed view perspectives to original ones across a desired viewing depth range.
• An end-to-end, result-oriented coding method using a depth-distributed fitting neural network paradigm was employed to render a fusion EIA.

Main Results:

• The proposed method successfully renders a fusion EIA with optimal multidepth fusion and resolution enhancement.
• High registration accuracies were achieved for both view perspective and depth reconstructions.
• Simulations and optical experiments demonstrated improved viewing resolution and an extended viewing depth range.

Conclusions:

• The developed neural network approach effectively enhances the performance of digital 3D displays.
• The method significantly improves viewing resolution and expands the depth of field (DoF).
• The feasibility of this approach for practical 3D display applications is verified.