EUVIMEDEUVIMED entdecken
Crossref

Rice Leaf Nitrogen Content Estimation Through A Methodological Framework Using Single-Sensor Multispectral Images

Muliady Ang, Tien Tze Lim, Voon Chet Koo, Jeremy Dimitri, Eric Chandra

Journal of Engineering Technology and Applied Physics · 15.09.2024 · 6 (2) · 38-46

DOI: 10.33093/jetap.2024.6.2.6 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Engineering Technology and Applied Physics, online erschienen am 15.09.2024. Verfasst von Muliady Ang und 4 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Using non-destructive evaluation tools based on imaging techniques, including single-sensor multispectral cameras, provides a cost-effective solution for optimizing rice nitrogen fertilization through site-specific nutrient management. However, their accuracy and precision have been identified as areas for improvement. This study aims to develop a methodology to improve the accuracy of estimations through field experiments. It utilizes multispectral images captured by MAPIR Survey3W Orange Cyan Near-Infrared and MAPIR Survey3W Red Edge cameras. The Normalized Difference Vegetation Index and Red Edge values derived from these images are correlated with Soil Plant Analysis Development values to assess rice nitrogen levels. A prediction model is then built using the Support Vector Regression algorithm. Findings from the experiments underscore the importance of addressing shadow effects, integrating the dataset on light intensity and image capture time, conducting radiometric calibration, filtering outlier data, employing image segmentation, and utilizing nonlinear Canova tests to enhance estimation accuracy. By configuring the Support Vector Regression model with RBF kernel, gamma set to 1.24, and epsilon set to 0.1, the R2 of the train data and validation data reaches 0.851, and 0.840 respectively. Meanwhile, the R2 of the test data achieves 0.793 with a mean absolute percentage error of 3.49 % and a root mean square error of 1.70. These findings underscore the potential of the proposed methodology to improve the estimation of rice nitrogen status based on single-sensor multispectral images, paving the way for more effective nutrient management strategies in rice cultivation.

Zitieren

APA

Ang, M., Lim, T. T., Koo, V. C., Dimitri, J., & Chandra, E. (2024). Rice Leaf Nitrogen Content Estimation Through A Methodological Framework Using Single-Sensor Multispectral Images. Journal of Engineering Technology and Applied Physics, 6(2), 38-46. https://doi.org/10.33093/jetap.2024.6.2.6

Vancouver

Ang M, Lim TT, Koo VC, Dimitri J, Chandra E. Rice Leaf Nitrogen Content Estimation Through A Methodological Framework Using Single-Sensor Multispectral Images. Journal of Engineering Technology and Applied Physics. 2024;6(2):38-46. doi:10.33093/jetap.2024.6.2.6

BibTeX
@article{Ang2024,
  author = {Ang, Muliady and Lim, Tien Tze and Koo, Voon Chet and Dimitri, Jeremy and Chandra, Eric},
  title = {Rice Leaf Nitrogen Content Estimation Through A Methodological Framework Using Single-Sensor Multispectral Images},
  journal = {Journal of Engineering Technology and Applied Physics},
  year = {2024},
  volume = {6},
  number = {2},
  pages = {38--46},
  doi = {10.33093/jetap.2024.6.2.6}
}
Kostenlos starten →