Collection of support scripts to read/write/process light field data
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
lib
README.md
convertAll.m
convertBlenderTo5D.m
getPointcloud.m

README.md

Usage

We assume that the convert*.m files as well as the lib/ directory are placed on the same level as the data. i.e. the folder containing convert*.m and lib/ also contains the folders additional/, stratified/, test/, and training/. You can run

convertAll.m 

to generate a LF.mat file for each of the scenes.

If you have only downloaded a subset of the dataset you can generate the LF.mat files using

convertBlenderTo5D('FOLDER')

The LF.mat includes the light field data (LF.LF), the ground truth (LF.depth/disp_high/lowres), a mask as well as the centerview. All parameters used for generation are loaded from the parameters.cfg and stored in LF.parameters. If you are familiar with the H matrix from Donald Dansereaus "Decoding, Calibration and Rectification for Lenselet-Based Plenoptic Cameras" (http://www-personal.acfr.usyd.edu.au/ddan1654/PlenCal.pdf) the provide the H matrix in LF.H. The distance between the two planes is stored in LF.f, given in [mm]. Please note that this is not the focal length of the respective cameras, which is stored in LF.parameters.intrinsics.focal_length_mm.

Generating point clouds

To generate point clouds you can either run

[X,Y,Z] = getPointcloud(LF)

or

[X,Y,Z] = getPointcloud(LF,'disp',d)

Here, LF is the variable as generated by the conversion scripts, i.e. it includes variables like LF, parameters and disp_lowres. If you do not specify a disparity map d, it will visualize the ground truth.

License

This work is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-sa/4.0/.

Authors: Katrin Honauer & Ole Johannsen

Website: www.lightfield-analysis.net

The 4D Light Field Benchmark was jointly created by the University of Konstanz and the HCI at Heidelberg University. If you use any part of the benchmark, please cite our paper "A dataset and evaluation methodology for depth estimation on 4D light fields". Thanks!

@inproceedings{honauer2016benchmark, title={A dataset and evaluation methodology for depth estimation on 4D light fields}, author={Honauer, Katrin and Johannsen, Ole and Kondermann, Daniel and Goldluecke, Bastian}, booktitle={Asian Conference on Computer Vision}, year={2016}, organization={Springer} }