TwGrey Algorithm Enhances Remote Sensing Image Classification While Reducing Feature Dimensions

Researchers Jingjing Qian, Shuhua Su, and Chujun Zhang have introduced TwGrey, an algorithmic feature selection framework designed to optimize remote sensing image classification by combining grey relational analysis and three-way decision theory. Published in the International Journal of Machine Learning and Cybernetics, the study addresses critical computational challenges in interpreting surface features across operational domains such as urban planning, disaster monitoring, and agriculture. The authors note that despite advances in classification models, feature selection remains a key step that affects model performance.

At the architectural core of TwGrey is the methodical integration of grey relational analysis with three-way decision theory. Within this structure, grey relational analysis is used to evaluate feature importance with low computational complexity. In parallel, three-way decision handles uncertainty by dividing features into positive, boundary, and negative regions.

An adaptive threshold calculation method based on shadowed set theory is proposed to reduce the reliance on manually specified thresholds.

The research team comprehensively validated TwGrey using established remote sensing benchmarks, conducting experiments on both the NWPU and UCM image datasets. When applied post-hoc to frozen Triplet-256 embeddings, TwGrey yielded consistent accuracy improvements over the full-dimensional baseline across all evaluated experimental settings. Notably, this is achieved while reducing the feature count by 38% to 79%, including on the full 45-class NWPU benchmark.

Statistical tests and stability analyses further indicate that the observed improvements are statistically significant in several settings.

Furthermore, TwGrey incurs higher computational latency than non-iterative filter methods. However, it achieves favorable speedups over comparable three-way wrapper baselines.

Subject of Research: Feature selection for remote sensing image classification using grey relational analysis and three-way decision theory

Article Title: Optimizing remote sensing image classification: a grey relational analysis and three-way decision approach

Article References: Qian, J., Su, S., & Zhang, C. (2026). Optimizing remote sensing image classification: a grey relational analysis and three-way decision approach. International Journal of Machine Learning and Cybernetics, 17(12), Article 536. https://doi.org/10.1007/s13042-026-03370-4

Image Credits: AI Generated

DOI: 10.1007/s13042-026-03370-4

Keywords: remote sensing image classification, TwGrey algorithm, feature selection, grey relational analysis, three-way decision theory, shadowed set theory, adaptive threshold calculation, NWPU dataset, UCM dataset, Triplet-256 embeddings, uncertainty regions, positive region, boundary region, negative region, three-way wrapper baselines, computational latency, surface feature interpretation, disaster monitoring