基于光谱图像的森林病虫害自动检测方法研究

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研究

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
2.
西安应用光学研究所，陕西西安 710065

基金项目: 兵器联合基金(6141B01020205)

详细信息

作者简介:
周晓丽（1978—），女，高级工程师，主要从事森林草原保护研究工作。E-mail：892851099@qq.com

中图分类号: TP39
计量
- 文章访问数: 35
- HTML全文浏览量: 30
- PDF下载量: 3
- 被引次数: 0
出版历程
- 网络出版日期: 2022-09-17

Research on automatic detection method of forest diseases and insect pests based on spectral image

1.
Forest and grassland protection and development center of Aruhorqin banner, Aruhorqin 025550, china
2.
Xi'an Institute of Appplied Optics, Xi'an 710065, china

摘要

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g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。
- 光谱图像 /
- 森林病虫害 /
- 深度学习 /
- 注意力机制
Abstract: The natural forest areas in China are widely distributed and the terrain is complex. Relying on the traditional patrol inspection method of forest rangers to prevent and control forest diseases and insect pests is inefficient, and it is difficult to find early forest diseases and insect pests in time. Therefore, the best time for prevention and control may be missed. In view of this problem, a deep learning network based on multispectral image detection of forest diseases and insect pests is designed, and a set of detection software is implemented. Through the UAV hanging flight experiment, the built deep learning network can complete the detection of infected areas in forest areas, and the detection results are analyzed.
- spectral image /
- Forest diseases and insect pests /
- deep learning /
- Attention mechanism

HTML全文

图 1 基于光谱图像的森林虫害检测的处理流程

Fig. 1 Processing flow of forest pest detection based on spectral image

下载: 全尺寸图片幻灯片

图 2 健康林木和染病林木相对光谱反射率曲线对比

Fig. 2 Comparison of relative spectral reflectance curves between healthy trees and infected trees

下载: 全尺寸图片幻灯片

图 3 深度学习检测网络结构

Fig. 3 Deep learning detection network structure

下载: 全尺寸图片幻灯片

图 4 金字塔池化模块内部结构

Fig. 4 Internal structure of pyramid pooling module

下载: 全尺寸图片幻灯片

图 5 注意力调整模块内部结构

Fig. 5 Internal structure of attention adjustment module

下载: 全尺寸图片幻灯片

图 6 检测软件界面

Fig. 6 Test software interface

下载: 全尺寸图片幻灯片

图 7 近距离染病区检测结果

Fig. 7 Detection results of close infected area

下载: 全尺寸图片幻灯片

图 8 远距离染病区检测结果

Fig. 8 Detection results of remote infected area

下载: 全尺寸图片幻灯片

表 1 不同尺度病害区域的检出率对比

Table 1 Comparison of detection rates in different scale disease areas

区域大小/像素	检出率/%
大于等于50×50	92.3
大于30×30小于50×50	65.6
小于等于30×30	23.1

下载: 导出CSV

表 2 本文方法与主流方法的比较

Table 2 Comparisons with state-of-the-art real-time methods

模型	图像分辨率/像素	MIoU/%	帧频/FPS
ENet ^[20]	1 024×512	53.1	60.5
ESPNet ^[21]	1 024×512	56.5	108.3
BiSeNetV2^[22]	1 024×512	65.5	152.1
BiSeNetV2-L^[22]	1 024×512	68.9	117
STDC1-Seg50^[23]	1 024×512	71.1	198.6
STDC2-Seg50^[23]	1 024×512	72.3	149.8
本文方法	1 024×512	75.2	137.1

下载: 导出CSV

表 3 消融实验情况

Table 3 Results of ablation experiments

金字塔池化模块	注意力调整模块	MIoU/%	帧频/FPS
√		70.3	156.6
	√	74.5	113.8
√	√	75.2	137.1

下载: 导出CSV

参考文献(23)

[1]	QIN Jun, WANG Biao, WU Yanlan, et al. Identifying pine wood nematode disease using UAV images and deep learning algorithms[J]. Remote Sensing,2021,13(2):162-168. doi: 10.3390/rs13020162
[2]	BECK P, ZARCOTEJADA P, STROBI P. The feasibility of detecting trees affected by the pine wood nematode using remote sensing[J]. Remote Sensing,2015(2):32-43.
[3]	YANG Hui, WU Penghai, YAO Xuedong, et al. Building extraction in very high resolution Imagery by dense-attention networks[J]. Remote Sensing,2018,10(11):1768-1784. doi: 10.3390/rs10111768
[4]	BADRINARAYANAN V, KENDALL A, CIPOLLA R. Segnet: A deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE transactions on pattern analysis and machine intelligence,2017,39(12):2481-2495. doi: 10.1109/TPAMI.2016.2644615
[5]	HUANG Zilong, WEI Yunchao, WANG Xinggang, et al. Alignseg: feature aligned segmentation networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2021,44(1):550-557.
[6]	FAHAD L, YASSINE R. Survey on semantic segmentation using deep learning techniques[J]. Neuro computing,2019,338:321-348.
[7]	CHEN L, PAPANDREOU G, KOKKINOS I, et al. DeepLab: semantic image segmentation with deep convolutional nets[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2018,40(4):834-848. doi: 10.1109/TPAMI.2017.2699184
[8]	MA Yanjun, YU Dianhai, WU Tian, et al. Paddlepaddle: an open-source deep learning platform from industrial practice[J]. Frontiers of Data and Domputing,2019(1):105-115.
[9]	GAO Kaijun, SUN Shaoyuan, YAO Guangshun, et al. Semantic segmentation of night vision images for unmanned vehicles based on deep learning[J]. Journal of Applied Optics,2017,38(3):421-428.
[10]	SHELHAMER E, LONG J, DARRELL T. Full convolutional networks for semantic segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(4):640-651. doi: 10.1109/TPAMI.2016.2572683
[11]	LIU S, CHEN J, RAHARD S. A new multi-focus image fusion algorithm and its efficient implementation[J]. IEEE Transactions on Circuit and Systems for Video Technology,2019(1):1-7.
[12]	SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for largle-scale image recognition[J]. arXiv preprint arxiv,2014(1):1409-1556.
[13]	ZHANG Xingchen. Deep learning-based multi-focus image fusion: a survey and a comparative study[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2012,44(1):4819-4838.
[14]	HAUHIGHAT M, AGHAGOLZADEH A, SEYEDARABI H. Muli-focus image fusion for visul sensor networks in DCT domain[J]. Computer & Electrical Engineering,2011,37(1):789-797.
[15]	SUJATHA K, PUNITHAVATHANI D. Optimized ensemble decision-based multi-focus image fusion using binary genetic grey-wolf optimizer in camera sensor networks[J]. Multimedia Tools Apply,2018,77(2):.1735-1759. doi: 10.1007/s11042-016-4312-3
[16]	LI Jinxing, GUO Xiaobao, Lu Guangming, et al. DRPL: deep regression pair learning for multi-focus image fusion[J]. IEEE Transactions on Image Processing,2020,29:4816-4831. doi: 10.1109/TIP.2020.2976190
[17]	DAI Min, DUAN Jinqiao, LIAO Junjun, et al. Maximum likelihood estimation of stochastic differential equations with random effects driven by fractional brownian motion[J]. Applied Mathematics and Computation,2021,397(5):1-13.
[18]	TAKENAKA Y, KATOH M, DENG S, et al. Detecting forests damaged by pine wood disease at the individual tree level using airborne laser data and worldview-2/3 images over two seasons[J]. ISPRS-International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences,2017(3):181-184.
[19]	JUNG H, KIM Y, JANG H, et al. Unsupervised deep image fusion with structure tensor representations[J]. IEEE Transactions on Image Processing,2022,29:3845-3858.
[20]	PASZKE A, CHAURASIA A, KIM S, et al. Enet: a deep neural network architecture for real-time semantic segmentation[J]. arXiv preprint arxiv,2016(3):6-7.
[21]	MEHTA S, RASTEGARI M, SHAPIRO L, et al. Espnetv2: a light-weight, power efficient, and general purpose convolutional neural network[J]. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition,2019(1):9190-9200.
[22]	YU Changqian, WANG Jingbo, PENG Chao, et al. Bisenet: bilateral segmentation network for real-time semantic segmentation[J]. In Proceedings of the European conference on computer vision (ECCV),2018(3):325-341.
[23]	FAN Mingyuan, LAI Shenqi, HUANG Junshi, et al. Rethinking bisenet for real-time semantic segmentation[J]. IEEE/CVF conference on computer vision and pattern recognition,2021(1):9716-9725.