Title: Calibration and Auto-Refinement for Light Field Cameras

Abstract: The ability to create an accurate three-dimensional reconstruction of a captured scene draws attention to the principles of light fields. This paper presents an approach for light field camera calibration and rectification, based on pairwise pattern-based parameters extraction. It is followed by a correspondence-based algorithm for camera parameters refinement from arbitrary scenes using the triangulation filter and nonlinear optimization. The effectiveness of our approach is validated on both real and synthetic data.