Polarimetric Helmholtz Stereopsis

1Louisiana State University, 2Tencent Pixel Lab, 3Zillow Group, 4George Mason University

Abstract


In this paper, we present the polarimetric Helmholtz stereopsis (polar-HS), which extends the classical Helmholtz stereopsis by considering the polarization state of light in the reciprocal paths. With the additional phase information from polarization, polar-HS requires only one reciprocal image pair. We formulate new reciprocity and diffuse/specular polarimetric constraints to recover surface depths and normals using an optimization framework. Using a hardware prototype, we show that our approach produces high-quality 3D reconstruction for different types of surfaces, ranging from diffuse to highly specular.

3D Reconstruction Results


Polarimetric Image Decomposition


Polarimetric Image Decomposition.

BibTeX

@ARTICLE {ding2024phs_pami,
  author    = {Ding, Yuqi and Ji, Yu and Cheng, Zhang and Zhou, Mingyuan and Kang, Sing Bing and Ye, Jinwei},
  title     = {Polarimetric Helmholtz Stereopsis},
  journal   = {IEEE Transactions on Pattern Analysis & Machine Intelligence},
  month     = {January},
  year      = {2024},
}
@InProceedings{ding2021phs_iccv,
  author    = {Ding, Yuqi and Ji, Yu and Zhou, Mingyuan and Kang, Sing Bing and Ye, Jinwei},
  title     = {Polarimetric Helmholtz Stereopsis},
  booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
  month     = {October},
  year      = {2021},
}