Calibrating an unfocused plenoptic camera based on parameters grouping and the light field structure point

Area of Science:

• Computer Vision
• Optical Engineering
• Image Processing

Background:

• Unfocused plenoptic cameras require precise calibration for reliable application.
• Existing projection models have limitations that hinder calibration accuracy.
• Advancements in light field imaging necessitate refined calibration techniques.

Purpose of the Study:

• To analyze the state-of-the-art plenoptic camera projection model.
• To propose a new projection model and a novel image feature representation called the light field structure point (LF-structure-point).
• To develop a simple and effective two-step calibration method utilizing the proposed LF-structure-point.

Main Methods:

• Detailed analysis of the current plenoptic camera projection model.
• Parameter grouping within the projection model.
• Introduction of the LF-structure-point for non-redundant light field data representation.
• Development of a 3D-to-3D projective transformation between light field structure space and real physical space.
• Algorithm for extracting LF-structure-points.

Main Results:

• Identification of parameter groups within the projection model.
• Proposal of a new projection model and the LF-structure-point.
• Demonstration that LF-structure-point offers a non-redundant signal structure representation.
• Successful implementation of a two-step calibration method using LF-structure-points.
• Satisfactory experimental validation of the proposed calibration method.

Conclusions:

• The proposed LF-structure-point provides an efficient and non-redundant representation of light field data.
• The novel projection model and calibration method simplify and enhance the accuracy of unfocused plenoptic camera calibration.
• This work contributes a significant advancement in light field camera calibration techniques.