Novel hybrid object-based non-parametric clustering approach for grouping similar objects in specific visual domains
- Submitting institution
-
Liverpool John Moores University
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 1013
- Type
- D - Journal article
- DOI
-
10.1016/j.asoc.2017.11.007
- Title of journal
- Applied Soft Computing
- Article number
- -
- First page
- 667
- Volume
- 62
- Issue
- -
- ISSN
- 1568-4946
- Open access status
- Compliant
- Month of publication
- November
- Year of publication
- 2017
- URL
-
-
- 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
-
1
- Research group(s)
-
-
- Citation count
- 6
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- The underpinning research was supported by a Scientific and Technological Research Council of Turkey award (Grant Number B.14.2.TBT.0.06.01-219-5-123). Handling variations in characteristics/distribution of large datasets/clusters is challenging for existing clustering approaches. This study focuses on the unsupervised object clustering specifically to deal with the dynamic number of clusters within multidimensional space in different visual domains. A novel clustering algorithm is introduced based on a pairwise similarity matrix to measure the optimum compromise between inter-class and intra-class similarity without prior knowledge. Tested on varying size datasets with diverse types of objects in specific visual domains, it produced robust outcomes (~90% accuracy).
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
