EUVIMEDEUVIMED entdecken
Crossref

Estimation of Leaf Area Index In Rice Using Combined Spectral and Texture Features From UAV Multispectral Imagery

Maria Victoria Bascon, Shunsaku Nishiuchi

The Philippine Journal of Crop Science · 31.08.2026 · 1

DOI: 10.63568/pjcs.v51i2 ↗

Kurz gefasst

Fachartikel in der Zeitschrift The Philippine Journal of Crop Science, online erschienen am 31.08.2026. Verfasst von Maria Victoria Bascon und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Leaf area index (LAI) is an important indicator of rice canopy development and productivity, yet accurately estimating it under field conditions remains challenging when relying solely on spectral vegetation indices. Because spectral reflectance captures radiometric properties of the canopy but not its spatial organization, a key question is whether supplementing spectral information with image-texture descriptors of canopy structure can yield more accurate and cultivar-robust LAI estimates. This study tested this proposition by combining UAV-derived spectral vegetation indices with gray-level co-occurrence matrix (GLCM) texture features for LAI estimation across five rice cultivars. Multispectral imagery and ground-measured LAI data were collected from field experiments conducted during the 2020 and 2021 rice growing seasons across five rice cultivars. Vegetation indices were derived from multispectral reflectance, while texture information was extracted using the GLCM feature and formulated into texture indices. Feature selection methods were applied in combination with multiple linear regression, ridge regression, support vector regression, and random forest. Among the evaluated approaches, random forest consistently produced the most stable LAI estimates (R2 = 0.60–0.65; RMSE = 0.68-0.73 m2 m-2). The inclusion of selected texture indices improved LAI estimation relative to vegetation indices alone, particularly when ratio-based vegetation indices were combined with near-infrared and red-band texture features. When models were evaluated separately for each cultivar, R² values ranged from 0.60 to 0.82, indicating that the spectral-texture approach accommodated variation in canopy architecture across cultivars. The observed improvements in estimation accuracy, particularly across cultivars with contrasting canopy architectures, suggest that spectral-texture integration can support more reliable UAV-based crop monitoring for cultivar-specific field management. However, generalizing these models across different cultivars and environmental conditions requires further investigation. Our current findings are based on data from typical growing seasons and selected cultivars and model performance on unseen varieties or atypical weather conditions remains uncertain. With these limitations, future research should explore genotype-environment interactions and work to enhance model robustness and adaptability. These results show that fusing texture information alongside spectral variables enhances UAV-based LAI estimation and provides a practical approach for field-scale rice canopy assessment.

Zitieren

APA

Bascon, M. V., & Nishiuchi, S. (2026). Estimation of Leaf Area Index In Rice Using Combined Spectral and Texture Features From UAV Multispectral Imagery. The Philippine Journal of Crop Science, 1. https://doi.org/10.63568/pjcs.v51i2

Vancouver

Bascon MV, Nishiuchi S. Estimation of Leaf Area Index In Rice Using Combined Spectral and Texture Features From UAV Multispectral Imagery. The Philippine Journal of Crop Science. 2026:1. doi:10.63568/pjcs.v51i2

BibTeX
@article{Bascon2026,
  author = {Bascon, Maria Victoria and Nishiuchi, Shunsaku},
  title = {Estimation of Leaf Area Index In Rice Using Combined Spectral and Texture Features From UAV Multispectral Imagery},
  journal = {The Philippine Journal of Crop Science},
  year = {2026},
  pages = {1},
  doi = {10.63568/pjcs.v51i2}
}
Kostenlos starten →