Cotton pest monitoring based on Logistic algorithm and remote sensing image
-
摘要:目的
借助多光谱遥感影像和Logistic算法，实现对棉田虫害的田间监测。
方法以患虫害棉花区域为研究对象，利用无人机获取棉田多光谱遥感影像，并对影像进行预处理；结合受虫害棉花光谱特征，利用虫害敏感波段反射率与植被指数构建Logistic回归模型，开展棉花虫害识别监测研究。
结果由土壤调节植被指数(Soil adjusted vegetation index，SAVI)模型和归一化植被指数(Normalized vegetation index，NDVI)模型构建的棉蚜虫、棉红蜘蛛、棉铃虫识别模型为最优模型，其训练样本准确率达到93.7%，测试样本准确率达到90.5%，召回率为96.6%，F1值为93.5%，对棉蚜虫、棉红蜘蛛和棉铃虫的识别模型决定系数分别为0.942、0.851和0.663。
结论该模型可满足棉田中棉蚜虫、棉红蜘蛛和棉铃虫3种虫害的发生区域识别，且可基本满足棉田精准植保作业相关要求。
Abstract:ObjectiveThe purpose of this article is to monitor cotton pest in field based on Logistic algorithms and multi-spectral remote sensing images.
MethodThe cotton areas with insect pests were selected as the research object. The multi-spectral remote sensing images of cotton field were acquired by UAV, and then pre-processed. Based on the spectral characteristics of cotton pests, the Logistic regression model was constructed by the reflectivity of pest-sensitive band and vegetation index to identify and monitor cotton pests.
ResultThe cotton aphid, cotton red spider mite, and cotton bollworm identification models constructed by the soil adjusted vegetation index (SAVI) model and the normalized vegetation index (NDVI) model were the optimal models, and their accuracy for training sample and test sample reached 93.7% and 90.5% respectively the recall rate and F1 value were 96.6% and 93.5% respectively and the determination coeffecients of recognition models for three types of pests were 0.942, 0.851 and 0.663 respectively.
ConclusionThis model can identify the occurrence area of cotton aphid, cotton red spider mite and cotton bollworm, which can basically meet the requirements of precision plant protection operation in cotton field.
-
-
![]()
图 4 辐射校正前（a）、后（b）单位像元内像素覆盖密度
Figure 4. Pixel coverage density in unit pixel before (a) and after radiation collection (b)
![]()
图 5 BP神经网络模型对辐射校正模型光谱反射率估测结果
Figure 5. Estimation results of spectral reflectance of image pixel after radiation correction by BP neural network model
![]()
图 7 健康棉叶及棉花虫害光谱波长反射率曲线
Figure 7. Spectral wavelength reflectance curves of healthy cotton leaves and cotton pests
![]()
图 10 3种棉田虫害受灾面积相关性分析
Figure 10. Correlation a Analysis of the areas affected by three types of pests in cotton field
表 1 绝对位置和方向不确定性参数
Table 1 Absolute geographic location and directional uncertainty parameters
指标
ItemX/m Y/m Z/m 偏转角/(°)
Deflection angle平移角/(°)
Translation angle扭转角/(°)
Twist angle均值 Average value 0.134 0.133 0.319 0.079 0.068 0.015 离散值(σ) Discrete value 0.022 0.026 0.067 0.009 0.014 0.002 
下载: 导出CSV
表 2 4种波段像素覆盖密度预测价值比较
Table 2 Comparison of the prediction values for the pixel coverage densities of four bands
辐射校正波段
Radiation correction band最佳截点
Best cut point举例数
No. of cases校正情况例数
No.of correction situation灵敏度/%
SensitivityROC曲线下面积
The area under the ROC curve好 Good 差 Bad 红光 Red <0.1 25 22 3 88.3 0.875 ≥0.1 225 199 26 蓝光 Blue 0 125 100 25 79.3 0.750 ≥0 125 100 25 近红外 Nir <0.5 13 11 2 80.9 0.688 ≥0.5 237 192 45 绿光 Green <0.5 63 55 8 87.1 0.875 ≥0.5 187 163 24
下载: 导出CSV
表 3 模型分类比较结果1）
Table 3 Comparison results of model classification
模型
Model实际样本
Actual
sample训练样本 Training sample 测试样本 Test sample 召回率/%
Recall rateF1 健康
HealthB1 B2 B3 总和
Sum准确率/%
Accuracy健康
HealthB1 B2 B3 总和
Sum准确率/%
AccuracyRMSE=0.193
n=1健康Health 50 3 1 1 55 91.4 28 1 1 2 32 85.8 93.9 89.8 B1 3 46 2 1 52 2 24 1 1 28 B2 2 0 46 0 48 1 1 18 1 21 B3 3 2 0 40 45 2 0 1 16 19 总和Sum 58 51 49 42 200 33 26 21 20 100 RMSE=0.187
n=2健康Health 50 1 1 1 53 93.7 25 1 0 0 26 90.5 96.6 93.5 B1 3 53 1 0 57 2 24 1 1 28 B2 3 0 45 0 48 2 1 29 1 33 B3 1 0 1 40 42 0 1 0 12 13 总和Sum 57 54 48 41 200 29 27 30 14 100 RMSE=0.376
n=3健康Health 67 6 5 2 80 83.1 22 3 3 2 30 75.1 90.4 82.8 B1 4 48 5 3 60 2 16 1 2 21 B2 3 1 25 1 30 2 2 20 2 26 B3 2 0 2 26 30 3 1 2 17 23 总和Sum 76 55 37 32 200 29 22 26 23 100 1) B1、B2和B3 分别表示棉蚜虫、棉红蜘蛛和棉铃虫
1) B1, B2 and B3 indicate cotton aphid, cotton red spider mite and cotton bollworm respectively
下载: 导出CSV
-
[1] 乔艳艳, 吴洁, 操宇林, 等. 不同药剂处理对棉种出苗及苗期病虫害发生的影响[J]. 安徽农业科学, 2020, 48(4): 125-127. doi: 10.3969/j.issn.0517-6611.2020.04.037 [2] 新疆农业科学院. 新疆棉花产量全国占比84.9%比重创新高[J]. 新疆农机化, 2020(1): 6. [3] 杜江涛. 棉花灌溉决策指标研究[J]. 农业科学, 2021, 11(4): 6. [4] 杨忠娜, 唐继军, 喻晓玲. 新疆棉花产业对国民经济的影响及对策研究[J]. 农业现代化研究, 2013, 34(3): 298-302. [5] 宋桂红. 新疆奎屯垦区机采棉品种引用、推广现状及对棉花品质的影响[J]. 中国种业, 2016(8): 44-45. [6] 王磊, 周建平, 许燕, 等. 农用无人机的应用现状与展望[J]. 农药, 2019, 58(9): 6-11. [7] 崔美娜, 戴建国, 王守会, 等. 基于无人机多光谱影像的棉叶螨识别方法[J]. 新疆农业科学, 2018(8): 1457-1466. [8] 兰玉彬, 邓小玲, 曾国亮. 无人机农业遥感在农作物病虫草害诊断应用研究进展[J]. 智慧农业, 2019, 1(2): 1-19. [9] 赵亮, 陈兵, 肖春华, 等. 棉花棉叶螨叶片遥感监测技术研究[J]. 甘肃农业大学学报, 2015(5): 94-99. doi: 10.3969/j.issn.1003-4315.2015.05.016 [10] PRABHAKAR M, PRASAD Y G, THIRUPATHI M, et al. Use of ground based hyperspectral remote sensing for detection of stress in cotton caused by leafhopper (Hemiptera: Cicadellidae)[J]. Computers & Electronics in Agriculture, 2011, 79(2): 189-198.
[11] 竞霞, 黄文江, 琚存勇, 等. 基于PLS算法的棉花黄萎病高空间分辨率遥感监测[J]. 农业工程学报, 2010(8): 229-235. doi: 10.3969/j.issn.1002-6819.2010.08.039 [12] 胡根生, 吴问天, 罗菊花, 等. 结合HJ卫星影像和最小二乘孪生支持向量机的小麦蚜虫遥感监测[J]. 浙江大学学报(农业与生命科学版), 2017(2): 211-219. [13] STELLA I R, GHOSH M. Modeling and analysis of plant disease with delay and logistic growth of insect vector[J]. Communications in Mathematical Biology and Neuroscience, 2018: 2018.
[14] 冯炼, 吴玮, 陈晓玲, 等. 基于HJ卫星CCD数据的冬小麦病虫害面积监测[J]. 农业工程学报, 2010, 26(7): 213-219. doi: 10.3969/j.issn.1002-6819.2010.07.037 [15] 马勇, 徐鑫, 徐文君, 等. Logistic方程模拟森林病虫害发生[J]. 辽宁林业科技, 2001(3): 17-18. doi: 10.3969/j.issn.1001-1714.2001.03.008 [16] MARCUS R. Deterministic and stochastic logistic models for describing increase of plant diseases[J]. 1991, 10(2): 155-159.
[17] YANG C, EVERITT J H, FERNANDEZ C J. Comparison of airborne multispectral and hyperspectral imagery for mapping cotton root rot[J]. Biosystems Engineering, 2010, 107(2): 131-139. doi: 10.1016/j.biosystemseng.2010.07.011
[18] 张连翔, 惠兴学, 赵树清. 植物病害I-S关系:Logistic衍生模型的研究[J]. 沈阳农业大学学报, 2001(4): 270-273. doi: 10.3969/j.issn.1000-1700.2001.04.008 [19] 谭宏卫, 曾捷. Logistic回归模型的影响分析[J]. 数理统计与管理, 2013(3): 476-485. [20] LI Y X, SHI H M. Research on data processing of low altitude photogrammetry based on Pix4D[J]. International Journal of Intelligent Information and Management Science, 2020, 9(1): 195-197.
[21] NISHAR A, RICHARDS S, BREEN D, et al. Thermal infrared imaging of geothermal environments and by an unmanned aerial vehicle (UAV): A case study of the Wairakei – Tauhara geothermal field, Taupo, New Zealand[J]. Renewable Energy, 2016, 86: 1256-1264. doi: 10.1016/j.renene.2015.09.042.
[22] 刘坦. Inpho、PhotoScan及Pix4D无人机正射影像处理软件对比[J]. 海峡科技与产业, 2017(11): 82-84. doi: 10.3969/j.issn.1006-3013.2017.11.031 [23] 刘世生, 王仁宗, 李叶民, 等. 一种基于卫星影像数据的新疆棉花区域识别方法及系统: CN111345214A[P]. 2020 - 06 - 30 [ 2020 - 03 - 04 ] . [24] 杨杰, 张莹莹, 王建雄, 等. 利用NDVI与EVI再合成的植被指数算法[J]. 遥感信息, 2020, 35(5): 131-137. [25] 邓江, 谷海斌, 王泽, 等. 基于无人机遥感的棉花主要生育时期地上生物量估算及验证[J]. 干旱地区农业研究, 2019, 37(5): 55-61. doi: 10.7606/j.issn.1000-7601.2019.05.09 [26] 唐亮, 何明珠, 许华, 等. 基于无人机低空遥感的荒漠植被覆盖度与归一化植被指数验证及其对水热梯度的响应[J]. 应用生态学报, 2020(1): 35-44. [27] 贾洁琼, 刘万青, 孟庆岩, 等. 基于GF-1 WFV影像和机器学习算法的玉米叶面积指数估算[J]. 中国图象图形学报, 2018, 23(5): 107-117.
计量
- 文章访问数: 2092
- HTML全文浏览量: 8
- PDF下载量: 730
