A Flexible Coding Scheme Based on Block Krylov Subspace Approximation for Light Field Displays with Stacked...

Area of Science:

• Computer Vision
• Image Processing
• Display Technology

Background:

• Realistic 3D perception on glasses-free displays requires continuous motion parallax, greater depth of field, and wider fields of view.
• Layered or Tensor light field 3D displays are gaining attention for their ability to support multiple simultaneous views at high resolution using minimal layers.

Purpose of the Study:

• To propose a novel, flexible scheme for efficient layer-based representation and lossy compression of light fields specifically for layered displays.
• To enhance the visual quality and reduce the data requirements for glasses-free 3D display applications.

Main Methods:

• A convolutional neural network (CNN) optimizes stacked multiplicative layers for light field representation.
• Block Krylov singular value decomposition (BK-SVD) analyzes low-rank structures to remove redundancy.
• High Efficiency Video Coding (HEVC) further compresses the approximated layer representation, eliminating inter-frame and intra-frame redundancies.

Main Results:

• The proposed scheme achieves significant bitrate savings compared to existing light field compression methods.
• It demonstrates flexibility in supporting multiple bitrates by adjusting BK-SVD ranks and HEVC quantization.
• The method integrates data-driven CNN approaches with efficient coding for practical display applications.

Conclusions:

• The developed scheme offers an efficient and flexible solution for light field compression on layered displays.
• It paves the way for improved glasses-free 3D display performance and broader adoption.
• The approach combines advanced signal processing techniques with machine learning for optimal compression.