用低成本传感器通过光反射差异识别柑橘黄龙病，适合农户早期筛查。
Low-Cost Optoelectronic Sensor for Early Screening of Citrus Greening in Leaves
- 基于白光与红外LED照射叶片，测量不同波段反射率变化。
- 红外波段识别准确率达89.58%，绿光波段达85.42%。
- 模型可部署于便携设备，适合小农户使用。
柑橘黄龙病（Huanglongbing, HLB）是一种严重危害柑橘作物的疾病，尚无治愈方法。早期检测至关重要，但现有方法成本较高。为此，本文开发了一种低成本、便携式传感器，利用基于LED的光学传感电路区分健康与感染HLB的柑橘叶片。该装置采用白光和红外（IR）LED照射叶片上表面，测量因健康与感染叶片生化成分差异导致的反射率变化。分析四个光谱波段（蓝、绿、红、红外）的数据，采用随机森林等机器学习模型进行处理。实验结果表明，红外波段表现最佳，随机森林模型准确率达89.58%，精确率为93.75%；绿光波段准确率为85.42%，精确率为90.62%。结果表明，该基于LED的光学系统可作为手持式筛查工具，为小规模农户提供经济高效的早期检测方案。
原文摘要 · Abstract (English)
Citrus greening, or Huanglongbing (HLB), is a serious disease affecting citrus crops, with no known cure. Early detection is essential, but current methods are often expensive. To address this, a low-cost, portable sensor was developed to distinguish between HLB-infected and healthy citrus leaves using a LED-based optical sensing circuit. The device uses white and infrared (IR) LEDs to illuminate the adaxial leaf surface and measures change in reflectance intensities caused by differences in biochemical compositions between healthy and HLB-infected leaves. These changes, analyzed across four spectral bands (blue, green, red, and IR), were processed using machine learning models, including Random Forest. Experimental results indicated that the IR band was the most effective, with the Random Forest model achieving an accuracy of 89.58% and precision of 93.75%. Similarly, the green band also achieved an accuracy of 85.42% and precision of 90.62%. These results suggest that this LED-based optical system could be a hand-held screening tool for early detection of HLB, providing small-scale farmers with a cost-effective solution.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。