多光譜影像技術用於早期檢測水稻蟲害之研究
Other Title
Preliminary Study on Multispectral Imaging Techniques for Early Pest Detection in Rice Production
Journal
台灣農業研究
Journal Volume
74
Journal Issue
3
Start Page
317
End Page
329
ISSN
2790-086X
Date Issued
2025-09-10
Author(s)
Abstract
本研究針對水稻主要害蟲褐飛蝨，結合人工接種試驗與多光譜影像技術，建立一套可應用於早期預警的蟲害評估模式。試驗中模擬不同褐飛蝨族群密度處理，並使用無人機搭載多光譜相機於田間進行影像監測，蒐集植株光譜資訊。所稱『早期』係指水稻自幼穗分化期後至抽穗期前之期間，約對應二期作水稻蟲害好發時段。分析結果顯示，當常態化差異植生指數 (normalized difference vegetation index; NDVI) 相對於對照組之差值超出自然變異範圍 (± 1s) 時，即可視為一種預警訊號，具有辨識褐飛蝨危害初期跡象之潛力，並可作為田間早期監測之初步依據。此外，本研究亦探討田區邊行水稻 (rice near border) 作為對照組替代方案之可行性，以提升實務應用之彈性。研究成果提供傳統農業人工監測以外之替代技術，亦為智慧農業蟲害監測提供可量化之操作依據。
This study focuses on the brown planthopper (Nilaparvata lugens (Stål)), a major pest in rice cultivation, and proposes an early-stage pest detection approach that integrates artificial infestation experiments with unmanned aerial vehicle-based multispectral imaging. Treatments with varying in-sect population densities were simulated, and rice spectral responses were monitored using multispec-tral imagery collected during the experiment. In this study, the term “early stage” refers to the period between the panicle initiation and heading stages, and corresponds to the typical outbreak period of brown planthopper (BPH) infestations in the second cropping season. Results showed that when the normalized difference vegetation index (NDVI) deviated from the control group by more than the nat-ural variation range (± 1s), such deviation could serve as an early-warning signal with the potential to indicate the onset of BPH infestation. The feasibility of using rice plants near field borders as a substi-tute reference for formal control plots was also evaluated to improve the feasibility for field operation. The proposed approach provides a quantifiable basis for pest monitoring and serves as a potential alternative to traditional manual scouting in the context of smart agriculture.
Subjects
褐飛蝨
蟲害監測
多光譜
智慧農業
Brown planthopper
Insect pest monitoring
Multispectral
Smart agriculture
Publisher
農業部農業試驗所
Type
journal article
File(s)![Thumbnail Image]()
Loading...
Name
74-3-9.pdf
Size
2.12 MB
Format
Adobe PDF
Checksum
(MD5):79d59587a1dd19d03375f592c686ce8a