| 基于光谱图像的森林病虫害自动检测方法 |
| |
| 引用本文： | 周晓丽,周立君,伊力塔,刘宇. 基于光谱图像的森林病虫害自动检测方法[J]. 应用光学, 2023, 44(2): 420-426. DOI: 10.5768/JAO202344.0203009 |
| |
| 作者姓名： | 周晓丽 周立君 伊力塔 刘宇 |
| |
| 作者单位： | 1.阿鲁科尔沁旗森林草原保护发展中心，内蒙古 阿鲁科尔沁旗 025550 |
| |
| 基金项目： | 兵器联合基金(6141B01020205) |
| |
| 摘 要： | 我国天然林区分布范围广，地形复杂，依靠传统的护林员巡检方式进行林木病虫害防治，效率较低，难于及时发现早期的林木病虫害，可能因此错过防治的最佳时机。针对该问题，设计了一种基于多光谱图像检测林木病虫害的深度学习网络，研发了一套检测软件，通过无人机挂飞实验，利用搭建的深度学习网络，完成林区染病区检测，对检测结果进行了分析。
|
| 关 键 词： | 光谱图像 森林病虫害 深度学习 注意力机制 |
| 收稿时间： | 2022-05-10 |
Automatic detection method of forest diseases and insect pests based on spectral images |
| |
| Zhou X.,Zhou L.,Yi L.,Liu Y.. Automatic detection method of forest diseases and insect pests based on spectral images[J]. Journal of Applied Optics, 2023, 44(2): 420-426. DOI: 10.5768/JAO202344.0203009 |
| |
| Authors: | Zhou X. Zhou L. Yi L. Liu Y. |
| |
| Affiliation: | 1.Forest and Grassland Protection and Development Center of Ar Horqin Banner, Ar Horqin Banner 025550, China2.Xi'an Institute of Applied Optics, Xi'an 710065, China |
| |
| Abstract: | The natural forest areas in China are widely distributed and the terrain is complex. Relying on the traditional patrol detection method of forest rangers to prevent and control forest diseases and insect pests is inefficient, so it is difficult to find early forest diseases and insect pests in time, which may miss the best time for prevention and control. In view of this problem, a deep learning network based on multispectral image detection of forest diseases and insect pests was designed, and a set of detection software was developed. Through the UAV hanging flight experiment, the built deep learning network was used to complete the detection of infected areas in forest areas, and the detection results were analyzed. |
| |
| Keywords: | attention mechanism deep learning forest diseases and insect pests spectral images |
| 本文献已被 维普 等数据库收录！ |
| 点击此处可从《应用光学》浏览原始摘要信息 |
|
点击此处可从《应用光学》下载免费的PDF全文 |
|
