A Monge–Ellipse Descriptor for Interpretable Quantification of Pore Architecture in Rock Images
Advanced Theory and Simulations · 2026-10 · 9 (10)
DOI: 10.1002/adts.70575 ↗Kurz gefasst
Fachartikel in der Zeitschrift Advanced Theory and Simulations, online erschienen am 08.10.2026. Verfasst von Raphael M. Tromer und 7 weiteren Autor:innen. Die Arbeit nennt 56 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Norges Forskningsråd, Fundação de Apoio à Pesquisa do Distrito Federal, Conselho Nacional de Desenvolvimento Científico e Tecnológico.
Abstract (Originalsprache)
Quantifying pore architecture in rocks is essential for understanding storage, transport, mechanical weakening, and reservoir quality. However, conventional porosity measurements reduce complex pore networks to a single scalar quantity. Here, we introduce a two‐dimensional Monge–Ellipse descriptor that combines pore fraction, pore abundance, pore‐size statistics, elliptical anisotropy, boundary morphology, and local inter‐pore organization into an interpretable morphological signature. The method adapts Monge's classical theorem by representing each pore through a second‐moment ellipse and computing direction‐dependent external homothetic centers for local pore triplets defined by Delaunay triangulation. The resulting Monge–Ellipse score measures departures from ideal homothetic organization and complements conventional shape descriptors. Applied to three segmented carbonate thin‐section images, the descriptor separates a low‐porosity, fragmented system; an intermediate‐porosity system with the strongest local Monge–Ellipse disorder; and a high‐porosity system dominated by larger connected pore regions. The most informative contrasts arise from upper‐tail statistics, particularly the 90th percentiles () of pore size, major axis, and Monge–Ellipse score, rather than from porosity or median shape alone. These results establish Monge‐inspired geometry as a physically interpretable route for transforming segmented rock images into feature vectors for quantitative petrography, image‐based rock typing, and machine‐learning studies of pore morphology.
Förderung
- Norges Forskningsråd
- Fundação de Apoio à Pesquisa do Distrito Federal
- Conselho Nacional de Desenvolvimento Científico e Tecnológico
Zitieren
APA
Tromer, R. M., Oliveira, I. D. S., Bastos, C. M. D. O., Dias, A. C., Barbosa, P. F., Vieira, L. C., Santos, R. V., & Junior, L. A. R. (2026). A Monge–Ellipse Descriptor for Interpretable Quantification of Pore Architecture in Rock Images. Advanced Theory and Simulations, 9(10). https://doi.org/10.1002/adts.70575
Vancouver
Tromer RM, Oliveira IDS, Bastos CMDO, Dias AC, Barbosa PF, Vieira LC, et al. A Monge–Ellipse Descriptor for Interpretable Quantification of Pore Architecture in Rock Images. Advanced Theory and Simulations. 2026;9(10). doi:10.1002/adts.70575
BibTeX
@article{Tromer2026,
author = {Tromer, Raphael M. and Oliveira, Israel da Silva and Bastos, Carlos Maciel de Oliveira and Dias, Alexandre Cavalheiro and Barbosa, Paola Ferreira and Vieira, Lucieth Cruz and Santos, Roberto Ventura and Junior, Luiz Antonio Ribeiro},
title = {A Monge–Ellipse Descriptor for Interpretable Quantification of Pore Architecture in Rock Images},
journal = {Advanced Theory and Simulations},
year = {2026},
volume = {9},
number = {10},
doi = {10.1002/adts.70575}
}
EUVIMED