Robust Depth and Reflectivity Estimation

Forward models describe what the sensor records. This thread is about the inverse problem: recovering scene properties from a handful of photons contaminated by ambient light.

Rank-ordered mean, revisited

The rank-ordered mean (ROM) estimator is a workhorse for depth estimation from sparse photon timestamps. (Yau et al., 2024) analyzes where it succeeds and where it fails, and proposes improvements that extend its usable range.

Joint depth and reflectivity

Depth and reflectivity are usually estimated separately, even though they are coupled through the same measurement. (Weerasooriya et al., 2025) estimates them jointly, using the structure of the coupling rather than fighting it.

References

2025

Preprint

Joint Depth and Reflectivity Estimation using Single-Photon LiDAR

Hashan K. Weerasooriya, Prateek Chennuri, Weijian Zhang, Istvan Gyongy, and Stanley H. Chan

arXiv preprint arXiv:2505.13250, Dec 2025

arXiv Bib HTML PDF

@article{weerasooriya2025joint,
  title = {Joint Depth and Reflectivity Estimation using Single-Photon LiDAR},
  author = {Weerasooriya, Hashan K. and Chennuri, Prateek and Zhang, Weijian and Gyongy, Istvan and Chan, Stanley H.},
  journal = {arXiv preprint arXiv:2505.13250},
  year = {2025},
}

2024

MMSP

Analysis and Improvement of Rank-Ordered Mean Algorithm in Single-Photon LiDAR

William C. Yau, Weijian Zhang, Hashan K. Weerasooriya, and Stanley H. Chan

In IEEE International Workshop on Multimedia Signal Processing (MMSP), Dec 2024

arXiv Bib HTML PDF

@inproceedings{yau2024analysis,
  title = {Analysis and Improvement of Rank-Ordered Mean Algorithm in Single-Photon LiDAR},
  author = {Yau, William C. and Zhang, Weijian and Weerasooriya, Hashan K. and Chan, Stanley H.},
  booktitle = {IEEE International Workshop on Multimedia Signal Processing (MMSP)},
  year = {2024},
}