Geometry parameter calibration for focused plenoptic cameras.

Area of Science:

• Optics and Photonics
• Computer Vision
• Image Processing

Background:

• Focused plenoptic cameras (FPCs) suffer from inaccurate geometry parameter retrieval post-packaging, leading to light field processing errors and visual artifacts.
• Precise calibration is crucial for high-fidelity light field rendering and applications.

Purpose of the Study:

• To propose a novel blind calibration method for accurately determining the geometry parameters of focused plenoptic cameras.
• To address the challenge of retrieving FPC parameters without prior knowledge of their exact configuration.

Main Methods:

• The method reframes geometry parameter calculation as determining micro-image patch size based on relay imaging geometry.
• A dark image calibration algorithm is used to find the microlens array (MLA) position and parameters.
• A triple-level calibration board with random textures facilitates blind focus plane confirmation and feature matching for patch size determination.

Main Results:

• The proposed Gradient-SSIM-based fractional-pixel matching accurately determines rendering patch size using geometry projection analysis.
• Experiments on simulated and real data validate the method's high accuracy and robustness across different FPCs.

Conclusions:

• The novel blind calibration method effectively retrieves FPC geometry parameters with high precision.
• This approach significantly enhances the accuracy of light field processing and reduces rendering artifacts in FPCs.