植物学报 ›› 2026, Vol. 61 ›› Issue (4): 0-0.

• •    下一篇

基于机器学习的水果成熟度无损检测技术研究进展

林卫国,李云川,陈先圣,周大猷,王大臣,王英力,黄远   

  1. 华中农业大学
  • 收稿日期:2025-12-13 修回日期:2026-01-28 出版日期:2026-07-10 发布日期:2026-05-28
  • 通讯作者: 王英力
  • 基金资助:
    国家自然科学基金;国家自然科学基金;中央高校基本科研业务费专项资金资助项目;中央高校基本科研业务费专项资金资助项目

Advances in Non-Destructive Testing Technology for Fruit Ripeness Based on Machine Learning

Xian-Sheng CHEN2, Ying-Li Wang3,远 黄2   

  • Received:2025-12-13 Revised:2026-01-28 Online:2026-07-10 Published:2026-05-28
  • Contact: Ying-Li Wang

摘要: 水果成熟度的无损检测对于降低水果生产、运输以及存储过程不必要的损失有重要意义。传统的水果成熟度检测方法主要包括经验判断法、有损检测法和无损检测法,其中经验判断法与有损检测法大都存在效率低下、准确性不足或导致样品破坏等问题。近年来,随着近红外光谱、高光谱成像、机器视觉、声学振动、电学特性、触觉感知及电子鼻等无损检测技术迅速发展,通过检测水果的颜色、硬度、糖酸比以及挥发性气体等与水果成熟度具有高度相关性的关键参数,实现了对水果成熟状态的精准评估。本研究以水果成熟过程中生理特性变化为切入点,系统梳理无损检测技术的基本原理与应用。此外,研究还总结了当前该领域面临的共性挑战并且还探讨了未来可能的发展方向。

关键词: 无损检测, 水果成熟度, 多模态融合, 传感器

Abstract: Non-destructive detection of fruit maturity is crucial for reducing unnecessary losses during fruit production, transportation, and storage. Traditional methods for assessing fruit maturity include empirical judgment, destructive testing, and non-destructive testing. Empirical judgment and destructive testing often suffer from low efficiency, inadequate accuracy, or cause sample damage.In recent years, non-destructive detection technologies, such as near-infrared spectroscopy, hyperspectral imaging, machine vision, fluorescence sensing, acoustic vibration, electrical properties, tactile sensing, and electronic noses, have rapidly developed. These technologies can precisely evaluate fruit maturity by measuring key parameters closely related to maturity, such as color, hardness, sugar-acid ratio, and volatile gases.This study reviews the principles of these non-destructive technologies and their current applications. It categorizes common fruit characteristics based on the core parameters that influence maturity, such as respiratory jump, color gradient, texture sensitivity, and aroma markers. It also summarizes the detection methods commonly used for different types of fruit.Additionally, the study identifies common challenges in this field, including significant environmental interference, difficulty in simultaneous multi-parameter detection, and lack of standardized systems. Finally, the study discusses potential future developments, including the integration of miniaturized multimodal sensors, the standardization of detection data, and the deeper integration of artificial intelligence with sensor systems.

Key words: Non-destructive testing, Fruit ripeness, multimodal fusion, Sensors

中图分类号: