Image-Based Relighting with 5-D Incident Light Fields

Abstract

In this paper, we propose a method for image-based relighting with 5-D incident light fields: 4 DoF of the position and direction and 1 DoF of the color of an incident ray. Specifically, we illuminate a scene with various rays by using a two-layer 5 DoF lighting system consisting of a rear-projection display and a transmissive LC panel, and synthesize images under desired 5-D incident light fields by combining the images captured under those rays. Our proposed method efficiently acquires the required images by using coded illumination; it reduces the number of captured images and the measurement time, and enhances their SNRs. In addition, we propose a method for removing the effects of the black offsets due to the projector and the LC panel in the two-layer setup. The experimental results using the prototype system show that our method enables us to synthesize photo-realistic images of scenes where wavelength-dependent phenomena such as fluorescence are observed.

Cite

Text

Oya and Okabe. "Image-Based Relighting with 5-D Incident Light Fields." IEEE/CVF International Conference on Computer Vision Workshops, 2017. doi:10.1109/ICCVW.2017.358

Markdown

[Oya and Okabe. "Image-Based Relighting with 5-D Incident Light Fields." IEEE/CVF International Conference on Computer Vision Workshops, 2017.](https://mlanthology.org/iccvw/2017/oya2017iccvw-imagebased/) doi:10.1109/ICCVW.2017.358

BibTeX

@inproceedings{oya2017iccvw-imagebased,
  title     = {{Image-Based Relighting with 5-D Incident Light Fields}},
  author    = {Oya, Shinnosuke and Okabe, Takahiro},
  booktitle = {IEEE/CVF International Conference on Computer Vision Workshops},
  year      = {2017},
  pages     = {3031-3038},
  doi       = {10.1109/ICCVW.2017.358},
  url       = {https://mlanthology.org/iccvw/2017/oya2017iccvw-imagebased/}
}