A family of globally optimal branch-and-bound algorithms for 2D–3D correspondence-free registration
- Submitting institution
-
Middlesex University
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 1226
- Type
- D - Journal article
- DOI
-
10.1016/j.patcog.2019.04.002
- Title of journal
- Pattern Recognition
- Article number
- -
- First page
- 36
- Volume
- 93
- Issue
- -
- ISSN
- 0031-3203
- Open access status
- Compliant
- Month of publication
- April
- Year of publication
- 2019
- URL
-
http://eprints.mdx.ac.uk/30210/
- 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
-
2
- Research group(s)
-
-
- Citation count
- 3
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- 2D/3D registration is crucial for forming visually accurate scene models for e.g. television/film production or architectural modelling. The underlying research problem is one of estimating the model pose in relation to the camera using a set of common features, which can become very complex and prone to inaccurate solutions when there are many poorly matching features. The significance of this paper lies in its entirely novel adaptation of 'branch-and-bound' techniques to match all scene primitives simultaneously. Crucially, this enables us, for the first time, to give concrete guarantees of a globally optimal solution.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -