Scholarlink

<p>This study presented a method integrating portable hyperspectral imaging and machine learning to enable non-destructive, in-field visualization of honey peach maturity. The effects of spectral preprocessing, feature selection, dimensionality reduction, and classification models on maturity discrimination performance were systematically investigated. Random Forest (RF) and XGBoost maturity classification models were constructed based on full spectral data, with their performance compared across different preprocessing methods. The RF model combined with wavelet transform preprocessing achieved 96.43% accuracy, while the XGBoost model performed best using first-derivative spectra (97.27%). Feature selection based on importance and PCA reduced the feature dimension to 20 key bands. Despite significant reduction in data complexity, classification accuracy remained above 95%. These key bands were primarily distributed in the visible pigment-sensitive, red-edge region and near-infrared regions associated with moisture and sugar content. They effectively reflected peel pigment transformation, tissue structural evolution, and internal physicochemical changes during honey peach maturation. Additionally, pixel-wise spatial visualization of maturity was realized based on the optimal classification model. In summary, this integrated approach enables rapid and non-destructive in-field assessment of honey peach maturity, with results provided as an intuitive spatial visualization. This approach provides critical technical support for intelligent and precision orchard harvesting.</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

In-field non-destructive maturity visualization of honey peach using portable hyperspectral imaging and machine learning

  • 期刊论文

  • 发表日期：

    2026年5月20日

查看全文
  • Hao Zhang,
  • Cheng Xie,
  • Cong Zhang,
  • Hongzhe Jiang,
  • Dachen Wang,
  • Hongping Zhou

摘要

This study presented a method integrating portable hyperspectral imaging and machine learning to enable non-destructive, in-field visualization of honey peach maturity. The effects of spectral preprocessing, feature selection, dimensionality reduction, and classification models on maturity discrimination performance were systematically investigated. Random Forest (RF) and XGBoost maturity classification models were constructed based on full spectral data, with their performance compared across different preprocessing methods. The RF model combined with wavelet transform preprocessing achieved 96.43% accuracy, while the XGBoost model performed best using first-derivative spectra (97.27%). Feature selection based on importance and PCA reduced the feature dimension to 20 key bands. Despite significant reduction in data complexity, classification accuracy remained above 95%. These key bands were primarily distributed in the visible pigment-sensitive, red-edge region and near-infrared regions associated with moisture and sugar content. They effectively reflected peel pigment transformation, tissue structural evolution, and internal physicochemical changes during honey peach maturation. Additionally, pixel-wise spatial visualization of maturity was realized based on the optimal classification model. In summary, this integrated approach enables rapid and non-destructive in-field assessment of honey peach maturity, with results provided as an intuitive spatial visualization. This approach provides critical technical support for intelligent and precision orchard harvesting.