Scholarlink

<p>Moisture content (MC) is a critical determinant of maize seed quality and storage stability. This study aimed to develop a rapid and non-destructive method for MC determination in single kernels using near-infrared spectroscopy (NIRS). To address the limitations of single-side measurement and unstable wavelength selection, we proposed a novel strategy integrating hybrid spectra (from both embryo and endosperm surfaces) and a stabilized variable selection algorithm (CARS-SPA). The CARS-SPA model, based on only seven optimal wavelengths, achieved superior predictive accuracy (R<sub>P</sub>=0.976 ± 0.004, RMSEP = 1.284 ± 0.021, RPD = 4.416 ± 0.072) with a high detection speed (260 ms/kernel). This research provides a robust framework for single-kernel moisture detection and facilitates the development of portable devices for industrial applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Near-infrared quantitative modelling of maize seed moisture based on hybrid spectroscopy and CARS-SPA coupling algorithm

  • 期刊论文

  • 发表日期：

    2025年9月27日

查看全文
  • Ren Zhang,
  • Jiaxin Wang,
  • Wei Wang,
  • Dinghai Xia,
  • Yingkun Bu,
  • Chaoqian Yang

摘要

Moisture content (MC) is a critical determinant of maize seed quality and storage stability. This study aimed to develop a rapid and non-destructive method for MC determination in single kernels using near-infrared spectroscopy (NIRS). To address the limitations of single-side measurement and unstable wavelength selection, we proposed a novel strategy integrating hybrid spectra (from both embryo and endosperm surfaces) and a stabilized variable selection algorithm (CARS-SPA). The CARS-SPA model, based on only seven optimal wavelengths, achieved superior predictive accuracy (RP=0.976 ± 0.004, RMSEP = 1.284 ± 0.021, RPD = 4.416 ± 0.072) with a high detection speed (260 ms/kernel). This research provides a robust framework for single-kernel moisture detection and facilitates the development of portable devices for industrial applications.