An entropy-based geometric feature fusion framework for automatic extraction of cartographic entities from digital map data
Accurate extraction of cartographic entities is essential for digital mapping and spatial data processing. This study proposes a geometric feature fusion framework to improve the identification of cartographic entities. Fourteen geometric descriptors were quantified and grouped into three categories: Visibility, Geometry, and Size. A two-stage fusion strategy was then developed, including initial fusion of individual metrics and deep fusion of the three categories using entropy-based weighting, to generate a unified significance representation. Based on this representation, an automatic extractor using the Classification and Regression Tree (CART) algorithm was constructed. Experiments on building datasets show that deeply fused factors outperform initially fused and pre-fusion factors, achieving a mean cross-validation accuracy of 99.89%, compared with 89.01% and 88.79%, respectively. The results demonstrate that geometric feature fusion is an effective and interpretable method for automated map generalisation.
Authors
- Zhe Li (ORCID: https://orcid.org/0009-0005-2898-8350)
- 彭慧
- Qingqing Deng
- Hongwei He
- Fei Zheng
- Binbin Ye
Institutions
- China Three Gorges University (CN)
- Gezhouba Group (China) (CN)
- Chengdu Surveying Geotechnical Research Institute (CN)
- Hubei Land Resources Vocational College (CN)
Publication Details
- Journal
- Geocarto International
- Published
- 2026-11-17
- DOI
- https://doi.org/10.1080/10106049.2026.2747512
- Primary Topic
- Automated Road and Building Extraction
- Type
- article
- Field-Weighted Citation Impact
- 0.00