High Quality Light Field Extraction and Post-Processing for Raw Plenoptic Data.

Area of Science:

• Computer Vision
• Optics
• Image Processing

Background:

• Light field technology is maturing, with applications in cinematography and virtual reality.
• Handheld plenoptic cameras offer flexibility but produce visual artefacts.
• Existing methods for sub-aperture image extraction have limitations.

Purpose of the Study:

• To propose and evaluate an advanced pipeline for enhancing light field images from handheld plenoptic cameras.
• To address visual artefacts and inconsistencies in captured light field data.
• To improve overall image quality for subsequent applications.

Main Methods:

• Demultiplexing RAW images to extract sub-aperture images.
• Implementing a pipeline involving hot pixel removal, color correction between views, and state-of-the-art light field denoising.
• Conducting an in-depth analysis of each pipeline step and their interactions.

Main Results:

• The proposed pipeline effectively removes hot pixel artefacts and corrects color inconsistencies.
• State-of-the-art light field denoising significantly improves image quality.
• The system demonstrates superior performance compared to existing sub-aperture image extraction algorithms based on quantitative metrics and subjective evaluation.

Conclusions:

• The developed pipeline offers a robust solution for enhancing light field data from handheld plenoptic cameras.
• The improved image quality positively impacts various light field applications, including cinematography and virtual reality.
• This work advances the practical usability of plenoptic camera technology.