shuozh/spinning-parallelogram-operator

Robust Depth Estimation for Light Field via Spinning Parallelogram Operator

Spinning-Parallelogram-Operator

Robust Depth Estimation for Light Field via Spinning Parallelogram Operator

CONTACT: Shuo Zhang
([email protected] or [email protected])

Any scientific work that makes use of our code should appropriately mention this in the text and cite our CVIU 2016 paper. For commercial use, please contact us.

PAPER TO CITE:

Shuo Zhang, Hao Sheng, Chao Li, Jun Zhang and Zhang Xiong.
Robust depth estimation for light field via spinning parallelogram operator, Computer Vision and Image Understanding, 2016, 145(C), 148-159

BIBTEX TO CITE:

@article{Zhang2016Robust,
  title={Robust depth estimation for light field via spinning parallelogram operator},
  author={Zhang, Shuo and Sheng, Hao and Li, Chao and Zhang, Jun and Xiong, Zhang},
  journal={Computer Vision and Image Understanding},
  volume={145},
  pages={148-159},
  year={2016},     }

Note:

This package also includes part of following softwares
Fast cost volume filtering

How to use:

Run demo.m
(This software is tested using Matlab 2013a with Windows 7 64bit environment)

input: Complete light field image, where x,y changes in the angular domain firstly and the spatial domain secondly.

The image pre-processing calculation method for CVIA dataset input is provided. Other example input images from HCI 4D Light Field Dataset and 4D Light Field Dataset (CVIA Konstanz & HCI Heidelberg) can be found at: https://drive.google.com/drive/folders/0B5JdDRk-RkPXTEJDdmhuRWUyVkU?usp=sharing

Time log:

2017.06.21 The package released.

2018.06.14 The InputImage.m is added.

spinning-parallelogram-operator's Issues

input/Buddha/lf.bmp

Are there any missing files?

SPO-MO

Any reference to SPO-MO?
Is it a follow-up of this work?

Questions about the evaluation scores on HCI 4D Light Field Benchmark

Dear authors of SPO algorithm,
Recently I've run your algorithm SPO on 8 scenes (stratified, training) offered by the HCI 4D light field benchmark, only to find the results completely different from those on the website, with all the parameters set as the website suggested. Then I realized that the conversion between pfm and png format can change the depth map contrast. For example, I tried to convert your results from pfm to png, ang again to pfm, then the scores are totally different. The result maps and scores, as well as the metric overviews are in the attachment. Could you please tell me how to handle this situation? Thank you very much, Looking forward to your reply.

How can I get opts.dmin or dmax of my own real LF image?

Hello, Can this work be used for real images? I have a light field device built by myself. How can I get opts.dmin?

您好，请教您一些问题

我用Lytro做实验的话，是要用解码出来的照片input对么？您有没有跑过这个代码呀