LiFMCR Dataset and Benchmark for Light Field Multi-Camera Registration

Aymeric Fleith*,1,2
Julian Zirbel*,1,2
Daniel Cremers1
Niclas Zeller2
*These authors contributed equally
1Technical University of Munich, Munich, Germany
2Karlsruhe University of Applied Sciences, Karlsruhe, Germany
Trajectories
Graphical representation of camera trajectories and frame poses for the two Raytrix R32 cameras in the main sequences of each scene.

Abstract

We present LiFMCR, a novel dataset for the registration of multiple MLA-based light field cameras. While existing light field datasets are limited to single-camera setups and typically lack external ground truth, LiFMCR provides synchronized image sequences from two high-resolution Raytrix R32 plenoptic cameras, together with high-precision 6-DOF poses recorded by a Vicon motion capture system. This unique combination enables rigorous evaluation of multi-camera light field registration methods.

As a baseline, we provide two complementary registration approaches: a robust 3D RANSAC-based method using cross-view point clouds, and a plenoptic PnP algorithm estimating extrinsic 6-DOF poses from single light field images. Both explicitly integrate the plenoptic camera model, enabling accurate and scalable multi-camera registration. Experiments show strong alignment with ground truth, supporting reliable multi-view light field processing.

Dataset

Dataset Scenes
Totally focused images processed from sample views of the scenes in the dataset.
Content of the LiFMCR dataset.
Scene Sequence type Number of frames
01_Plants Random camera movements 323
Handheld movements around the scene 632
02_Bike Random camera movements 434
Handheld movements around the scene 781
Movements in x, y, z directions 250
03_Office Random camera movements 328
Handheld movements around the scene 688
Fast movements 247
04_Electronics Random camera movements 323
Handheld movements around the scene 185
Random movements with lower exposure 428
05_Oscilloscope Random camera movements 406
06_Skeleton Random camera movements 533
Handheld close movements 758
Cameras in circle 677
07_Tools Random camera movements 471

Pipeline

Pipeline
Pipeline for registering camera images and estimating 6-DOF extrinsics between views. Camera 0 serves as reference and other cameras are registered using one of the two proposed methods. Note that the point clouds from cameras 1 to X are required only for the RANSAC method, not for the plenoptic PnP algorithm.
3DRANSAC PnP
Pipeline of the two proposed camera registration algorithms: the method left uses 3D pose estimation via RANSAC, and the method right uses PnP.

License Terms

LiFMCR was developed in collaboration between the Technical University of Munich and the Karlsruhe University of Applied Sciences. The code is open-source under a GNU General Public License Version 3 (GPLv3).

BibTeX

@inproceedings{FleithZirbel2025LiFMCR,
        title     = {LiFMCR: Dataset and Benchmark for Light Field Multi-Camera Registration},
        author    = {Fleith, Aymeric and Zirbel, Julian and Cremers, Daniel and Zeller, Niclas},
        booktitle = {International Symposium on Visual Computing (ISVC)},
        year      = {2025},
    }