A Novel Light Field Image Compression Method Using EPI Restoration Neural Network

Area of Science:

• Computer Vision
• Image Processing
• Data Compression

Background:

• Light field images capture both spatial and angular information, leading to large data volumes.
• Efficient storage and compression of light field data are significant challenges.
• Epipolar plane images (EPIs) possess low-rank properties suitable for data recovery.

Purpose of the Study:

• To propose a novel light field image coding framework.
• To improve compression performance and robustness for light field data.
• To utilize the inherent similarities within light field images.

Main Methods:

• Development of a light field image coding framework.
• Integration of an epipolar plane image (EPI) restoration neural network.
• Exploitation of inherent similarities in light field images.

Main Results:

• The proposed framework demonstrates superior performance compared to state-of-the-art methods.
• Quantitative and qualitative experimental results validate the method's effectiveness.
• The approach shows enhanced robustness in light field image compression.

Conclusions:

• The proposed EPI restoration neural network-based framework offers advanced light field image compression.
• The method effectively leverages light field data characteristics for improved efficiency.
• This technique presents a significant advancement in light field data handling.