Per-pixel distortion correction method of light field camera with homography matrix and phase calculation

Abstract

A light field camera is constructed by positioning a microlens array between the main lens and an imaging plane, allowing it to record depth information. One of the factors affecting the reconstruction accuracy is distortion. To address this, we propose a novel method for correcting distortion in light field cameras using a homography matrix and phase calculation of the fringe projection technique. In this method, the distortion characteristic of each pixel in light field cameras can be directly measured. Firstly, the phase values of the concentric circular target captured by the light field camera are acquired using phase calculation technique. Next, the homography matrix is obtained by extracting the center coordinates of the target and their corresponding phase values. The phase value of each ideal point is obtained using the homography matrix, whereas the phase value of the corresponding real point is obtained through phase calculation technique. Finally, the distortion mapping table is generated based on the per-pixel ideal and real point pairs. By comparing the full-field error of the images before and after correction, the results show an average improvement of 15.992 % using this method.