| 多光谱技术在果树病害检测中的应用与展望 |
| 点此下载全文 |
| 引用本文：岳柳羊,何雄奎,苏立阳,陈恒,王惟实,刘亚佳.多光谱技术在果树病害检测中的应用与展望.植物保护学报,2026,53(1):95-110 |
| DOI：10.13802/j.cnki.zwbhxb.2026.2026807 |
| 摘要点击次数: |
| 全文下载次数: |
|
|
| 中文摘要:多光谱技术作为计算机视觉与农业遥感交叉领域的核心技术，正在推动果园病虫害检测手段的革新与精准管理的实施。近年来，该技术与深度学习模型相结合，在果树病虫害的精准识别方面取得显著进展，并在性价比、适用性和实时监测等方面展现出独特优势。该文系统综述了多光谱技术在不同果树病害检测中的具体应用，例如，基于520~920 nm波段的多光谱遥感技术实现了梨树火疫病的检测，检测精度达95.0%；无人机多光谱影像在475、560、668、717和840 nm波段上对栗树油墨病的检测精度最高达95.2%；此外，融合多色荧光与反射波段的多光谱成像技术对柑橘黄龙病的检测精度达92.1%。结合支持向量机、随机森林及改进的Mask R-CNN V3等模型，多光谱技术在多种病害识别中进一步提高了检测精度与效率。支持向量机模型对野生蓝莓病害的检测精度达到96.60%；随机森林模型对槟榔黄化病的检测精度为86.46%；改进的Mask R-CNN V3模型对柑橘黄龙病的检测精度达到93.37%。另外，多光谱技术通过获取作物在多波段下的光谱信息，可有效反映叶片色素含量等生理状态，为病害早期诊断提供依据。未来，多光谱技术可通过融合机器学习算法增强模型泛化能力，并结合小型传感器与嵌入式计算平台，开发轻量化实时检测设备，实现果园病害的早期预警与精准防控，以期为智慧农业提供重要技术支持。 |
| 中文关键词:多光谱技术 果园病害检测 数据处理 深度学习 |
| |
| Application and prospects of multispectral technology in fruit tree disease detection |
| Author Name | Affiliation | E-mail | | Yue Liuyang | College of Science, China Agricultural University, Beijing 100193, China | | | He Xiongkui | College of Science, China Agricultural University, Beijing 100193, China | xiongkui@cau.edu.cn | | Su Liyang | College of Science, China Agricultural University, Beijing 100193, China | | | Chen Heng | College of Science, China Agricultural University, Beijing 100193, China | | | Wang Weishi | College of Science, China Agricultural University, Beijing 100193, China | | | Liu Yajia | College of Science, China Agricultural University, Beijing 100193, China | liuyajia@cau.edu.cn |
|
|
| Abstract:Multispectral technology, as a core technology at the intersection of computer vision and agricultural remote sensing, is driving innovation in orchard pest and disease detection methods and the implementation of precision management. In recent years, the integration of this technology with deep learning models has led to significant progress in the accurate identification of tree diseases and pests, demonstrating distinct advantages in cost-effectiveness, applicability, and real-time monitoring. This paper systematically reviews the applications of multispectral technology in the detection of various tree diseases. For example, multispectral remote sensing in the 520-920 nm wavelength range has been used to detect fire blight in pear trees, achieving a detection accuracy of 95.0%; unmanned aerial vehicle (UAV)-based multispectral imagery in the 475, 560, 668, 717, and 840 nm bands has achieved a maximum detection accuracy of 95.2% for ink disease in chestnut trees; furthermore, multispectral imaging technology integrating multi-color fluorescence and reflectance bands has achieved a detection accuracy of 92.1% for citrus huanglongbing. When combined with models such as Support Vector Machines (SVM), Random Forest, and the improved Mask R-CNN V3, multispectral technology has further enhanced detection accuracy and efficiency across multiple disease types. The SVM model achieved a detection accuracy of 96.60% in wild blueberry disease classification; the Random Forest model achieved an accuracy of 86.46% in detecting yellow leaf disease in betel nut trees; and the improved Mask R-CNN V3 model achieved a detection accuracy of 93.37% for citrus huanglongbing. In addition, by capturing spectral information across multiple bands, multispectral technology can effectively reflect physiological states such as leaf pigment content, providing a scientific basis for early disease diagnosis. In the future, this technology may further enhance model generalization through integration with machine learning algorithms and, in combination with miniaturized sensors and embedded computing platforms, enable the development of lightweight real-time detection devices for early warning and precise control of orchard diseases, thereby providing important technical support for smart agriculture. |
| keywords:multispectral technology orchard disease detection data processing deep learning |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
|
您是本站第 16177881 版权所有：植物保护学报 京ICP备05006550号-2 主管单位：中国科协 主办单位：中国植物保护学会、中国农业大学 地址：北京市圆明园西路2号 中国农业大学植物保护学院 植物保护学报编辑部 电话：010-62732528 电子邮件：zbxb@cau.edu.cn 技术支持： 北京勤云科技发展有限公司 |
