A Linear Fractional Transformation Model and Calibration Method for Light Field Camera

TL;DR

This work addresses the challenge of calibrating light field cameras by introducing a Linear Fractional Transformation (LFT) model with parameter $α$ that decouples the main lens from the MLA, enabling separate calibration and accelerated data generation for data-driven methods. An MLA descriptor matrix $H_α$ is introduced, together with an analytical least-squares solution and subsequent nonlinear refinement to estimate intrinsic and MLA parameters, alongside a CIP-based feature detection strategy. The method is validated on physical and simulated data, showing sub-pixel reprojection accuracy for the light-field camera and a faster, GPU-free simulator that outperforms prior tools like PRISM in runtime. The approach yields a robust, physically interpretable framework for calibration and rapid light-field image synthesis, with open-source code to support adoption and further research.

Abstract

Accurate calibration of internal parameters is a crucial yet challenging prerequisite for 3D reconstruction using light field cameras. In this paper, we propose a linear fractional transformation(LFT) parameter $α$ to decoupled the main lens and micro lens array (MLA). The proposed method includes an analytical solution based on least squares, followed by nonlinear refinement. The method for detecting features from the raw images is also introduced. Experimental results on both physical and simulated data have verified the performance of proposed method. Based on proposed model, the simulation of raw light field images becomes faster, which is crucial for data-driven deep learning methods. The corresponding code can be obtained from the author's website.

Figures (14)

• Figure 1: Decoupled projection model of light field camera. (a) for the main lens. (b) for the MLA.
• Figure 2: (a) One typical de-focused light field image of corner. (b) LSD result of (a).
• Figure 3: The result of filtering, the red points are the centers of micro lens images that satisfy the texture-richness condition.
• Figure 4: (a) One typical de-focused light field image of corner. (b) Line Segment Detector (LSD) result corresponding to (a). (c) Sampling result of (a) with the four quadrants highlighted in different colors. (d) A typical sampling result from a non-corner micro lens image.
• Figure 5: The visualization of the correspondence between 3D spatial points and their associated micro lens clusters.