Parallel and scalable heat methods for geodesic distance computation
- Submitting institution
-
Cardiff University / Prifysgol Caerdydd
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 96927005
- Type
- D - Journal article
- DOI
-
10.1109/TPAMI.2019.2933209
- Title of journal
- IEEE Transactions on Pattern Analysis and Machine Intelligence
- Article number
- -
- First page
- 579
- Volume
- 43
- Issue
- 2
- ISSN
- 0162-8828
- Open access status
- Compliant
- Month of publication
- August
- Year of publication
- 2019
- URL
-
https://doi.org/10.1109/TPAMI.2019.2933209
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
5
- Research group(s)
-
V - Visual computing
- Citation count
- 0
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- The classical "heat method" is a popular approach for computing geodesic distance but requires solving linear systems, making them unsuitable for very large models due to high memory consumption and/or high computational cost. We reformulated the heat method as an optimisation problem, and developed a parallel solver that involves no linear system. Our solver can handle meshes with more than 200 million vertices on a PC with 128GB RAM, while the classical heat method fails on such meshes. An open-source implementation has been released at https://github.com/bldeng/ParaHeat.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -