5种烟草常用农药的高光谱识别技术研究

doi:10.13304/j.nykjdb.2024.0458

摘要/Abstract

摘要：

为快速、准确获取烟叶生产田间用药类型，提升烟叶田间用药管控科学性和针对性，应用高光谱成像技术，通过对比分别喷施5种烟草常用农药48 h后的鲜烟叶光谱曲线差异，测试不同光谱预处理方法、特征波长提取方法和模式识别方法组合下的模型准确率。结果表明，在750~875 nm波段，5种农药处理后的鲜烟叶光谱曲线反射率存在差异。全波段下，标准正态变换和最小二乘支持向量机模型组合以及二阶导数和随机森林模型组合均具有很高的识别准确率，测试集准确率均达到98.58%；连续投影算法降维效果优于竞争性自适应重加权采样算法。特征波段下，二阶导数、连续投影算法和随机森林的模型组合表现最优，训练集识别准确率为100.00%，测试集识别准确率为98.22%，特征波段数量为17，单样本检测时间为10.28 ms，该方法可快速、准确地识别5种烟草生产常用农药类型。研究结果为烟叶生产田间用药管控提供技术支持。

关键词: 烟叶生产, 高光谱成像, 农药识别, 随机森林

Abstract:

To rapidly and accurately obtain the pesticide types used in tobacco field production， and to improve the scientific and targeted management of tobacco field pesticide applications， hyperspectral imaging technology was applied. By comparing the spectral curves of fresh tobacco leaves 48 h after spraying with 5 common pesticides， the spectral differences were analyzed. Various combinations of spectral preprocessing methods， feature wavelength extraction methods， and pattern recognition techniques were tested to evaluate model accuracy. The results showed that， in the 750~875 nm spectral range， the spectral curves of fresh tobacco leaves treated with the 5 pesticides exhibited distinct reflectance differences. Across the full spectral range， the combination of standard normal variate （SNV） transformation and least squares support vector machine （LSSVM） models， as well as the combination of second derivative preprocessing and random forest models， both achieved high recognition accuracy， with the test set accuracy reaching 98.58%. The continuous projection algorithm outperformed the competitive adaptive reweighting sampling algorithm in dimensionality reduction. In the feature wavelength range， the combination of second derivative preprocessing， continuous projection algorithm， and random forest models achieved the best performance， with the training set accuracy reaching 100.00% and the test set accuracy at 98.22%. The number of feature wavelengths was 17， and the single-sample detection time was 10.28 ms. This method could rapidly and accurately identify the types of 5 common pesticides used in tobacco production. Above results provided technical support for the management of pesticide application in tobacco field production.

Key words: tobacco field production, hyperspectral imaging, pesticide identification, random forest

中图分类号:

S220

杨国涛, 张世杰, 陈超, 刘云, 贺琛, 宁英豪, 张勍. 5种烟草常用农药的高光谱识别技术研究[J]. 中国农业科技导报, 2025, 27(7): 122-132.

Guotao YANG, Shijie ZHANG, Chao CHEN, Yun LIU, Chen HE, Yinghao NING, Qing ZHANG. Identification of 5 Common Pesticides Used in Flue-tobacco Field Production Based on Hyperspectral Technology[J]. Journal of Agricultural Science and Technology, 2025, 27(7): 122-132.

图/表 17

参考文献 31

[1]	司晓喜,刘志华,李中昌,等.5种农药在采收前和初烤后的烟叶及土壤中的残留量变化[J].烟草科技,2016,49(7):44-50.
	SI X X, LIU Z H, LI Z C, et al.. Residues of five pesticides in tobacco leaves before harvesting and after flue-curing and soils [J]. Tob. Sci. Technol., 2016, 49(7): 44-50.
[2]	罗华元,王绍坤,常寿荣,等.12种农药在烟叶中残留及烟气中转移试验初报[J].云南农业大学学报(自然科学版),2010,25(5): 636-641.
	LUO H Y, WANG S K, CHANG S R, et al.. Determination of 12 pesticides residues in tobacco leaf and shifting to smoke [J].J. Yunnan Agric. Univ. (Nat. Sci.), 2010, 25(5): 636-641.
[3]	王刚,孔凡玉,刘建利,等. 烟草田间农药合理使用规程: [S].北京:中国标准出版社,2011.
[4]	熊巍,张海燕,靳冬梅,等.UPLC-MS/MS同时测定烟草中5种新烟碱杀虫剂残留[J]. 湖南农业科学, 2021(2): 91-95.
	XIONG W, ZHANG H Y, JIN D M, et al.. Determination of five neonicotinoid insecticide residues in tobacco by ultra performance liquid chromatography tandem mass spectrometry [J]. Hunan Agric. Sci., 2021(2): 91-95.
[5]	许泳吉,刘云云,韩玮,等.固相萃取-高效液相色谱法同时测定烟草中吡虫啉和多菌灵[J].青岛科技大学学报(自然科学版), 2016, 37(1): 26-31.
	XU Y J, LIU Y Y, HAN W, et al.. Simultaneous detection of carbendazim and imidacloprid in tobacco with solid-phase extraction and high-performance liquid chromatography [J]. J. Qingdao Univ. Sci. Technol. (Nat. Sci.), 2016, 37(1): 26-31.
[6]	王彧婕, 冯德建, 冯双, 等. 吡虫啉和啶虫脒胶体金免疫层析法试纸条性能评估及应用效果探究[J].食品安全质量检测学报, 2024, 15(3): 133-139.
	WANG Y J, FENG D J, FENG S, et al.. Performance evaluation and application effect study of colloidal gold immunochromatographic test strips for imidacloprid and acetamiprid [J]. J. Food Saf. Qual., 2024, 15(3): 133-139.
[7]	楼小华,高川川,朱文静,等.胶体金免疫层析法快速检测烟草中吡虫啉残留[J].食品安全质量检测学报,2017,8(5):1739-1744.
	LOU X H, GAO C C, ZHU W J, et al.. Rapid detection of imidacloprid residues in tobacco leaf by colloidal gold immunochromatographic assay [J]. J.Food Saf. Qual., 2017, 8(5): 1739-1744.
[8]	孙杰. 基于荧光高光谱技术的红茶农药残留检测[D]. 成都:四川农业大学, 2023.
	SUN J. Detection of pesticide residues in black tea based on fluorescence hyperspectral technology [D]. Chengdu: Sichuan Agricultural University, 2023.
[9]	SUN J, CONG S L, MAO H P,et al..Quantitative detection of mixed pesticide residue of lettuce leaves based on hyperspectral technique [J/OL].J.Food Process Eng., 2018, 41(2): 12654 [2024-05-16]. .
[10]	吉海彦,任占奇,饶震红.高光谱成像技术鉴别菠菜叶片农药残留种类[J].发光学报,2018,39(12):1778-1784.
	JI H Y, REN Z Q, RAO Z H. Identification of pesticide residue types in spinach leaves based on hyperspectral imaging [J].Chin. J. Lumin., 2018, 39(12):1778-1784.
[11]	LU X C, ZHAO C, QIN Y Q, et al.. The application of hyperspectral images in the classification of fresh leaves' maturity for flue-curing tobacco [J/OL]. Processes, 2023, 11(4):1249 [2024-05-06]. .
[12]	JIA F F, HAN S, CHANG D, et al.. Monitoring flue-cured tobacco leaf chlorophyll content under different light qualities by hyperspectral reflectance [J]. Am. J. Plant Sci., 2020, 11(8): 1217-1234.
[13]	潘兆杰,孙根云,张爱竹,等.基于波段选择的烟草病害检测模型[J].光谱学与光谱分析,2023,43(4):1023-1029.
	PAN Z J, SUN G Y, ZHANG A Z, et al.. Tobacco disease detection model based on band selection [J]. Spectrosc. Spectr. Anal., 2023, 43(4): 1023-1029.
[14]	张慧,张文伟,张永毅,等.基于高光谱与纹理融合的烤烟分类方法研究[J].中国烟草学报,2022,28(3):72-80.
	ZHANG H, ZHANG W W, ZHANG Y Y,et al.. Research on classification method of flue-cured tobacco based on fusion of hyperspectral and texture features [J]. Acta Tab. Sin., 2022, 28(3): 72-80.
[15]	MARK H. Use of Mahalanobis distances to evaluate sample preparation methods for near-infrared reflectance analysis [J].Anal. Chem., 2002, 59(5): 790-795.
[16]	GALVÃO R K H, ARAUJO M C U, JOSÉ G E, et al.. A method for calibration and validation subset partitioning [J]. Talanta, 2005, 67(4): 736-740.
[17]	RINNAN Å, VAN DEN BERG F, ENGELSEN S B. Review of the most common pre-processing techniques for near-infrared spectra [J]. Trends Anal. Chem., 2009, 28(10): 1201-1222.
[18]	ENGEL J, GERRETZEN J, SZYMANSKA E, et al.. Breaking with trends in pre-processing [J]. Trends Anal. Chem., 2013, 50: 96-106.
[19]	RINNAN Å. Pre-processing in vibrational spectroscopy-when, why and how [J]. Anal. Meth., 2014, 6(18): 7124-7129.
[20]	LEE L C, LIONG C Y, JEMAIN A A. A contemporary review on data preprocessing (DP) practice strategy in ATR-FTIR spectrum [J]. Chemom. Intell. Lab. Syst., 2017, 163: 64-75.
[21]	WAN X H, LI G, ZHANG M Q, et al.. A review on the strategies for reducing the non-linearity caused by scattering on spectrochemical quantitative analysis of complex solutions [J]. Appl. Spectro. Rev., 2020, 55(5): 351-377.
[22]	唐杰,罗彦波,李翔宇,等.基于近红外光谱数据的一维卷积神经网络模型研究[J].光谱学与光谱分析, 2024, 44(3): 731-736.
	TANG J, LUO Y B, LI X Y, et al.. Study on one-dimensional convolutional neural network model based on near-infrared spectroscopy data [J]. Spectro. Spectr. Anal., 2024, 44(3):731-736.
[23]	李国厚,李泽旭,金松林,等.混合卷积神经网络用于高光谱小麦品种鉴别[J].光谱学与光谱分析,2024,44(3):807-813.
	LI G H, LI Z X, JIN S L, et al.. Mix convolutional neural networks for hyperspectral wheat variety discrimination [J].Spectro. Spectr. Anal., 2024, 44(3): 807-813.
[24]	KE Y C, SHI Z K, LI P J, et al.. Lithological classification and analysis using Hyperion hyperspectral data and random forest method [J]. Acta Petrol. Sin., 2018, 34(7): 2181-2188.
[25]	ARAÚJO M C U, SALDANHA T C B, GALVÃO R K H, et al.. The successive projections algorithm for variable selection in spectroscopic multicomponent analysis [J]. Chemom. Intell. Lab. Syst., 2001, 57(2): 65-73.
[26]	MOREIRA E D T, PONTES M J C, GALVÃO R K H, et al.. Near infrared reflectance spectrometry classification of cigarettes using the successive projections algorithm for variable selection [J]. Talanta, 2009, 79(5): 1260-1264.
[27]	SOARES S F C, GOMES A A, ARAUJO M C U, et al.. The successive projections algorithm [J]. Trends Anal. Chem., 2013, 42: 84-98.
[28]	LI H D, LIANG Y Z, XU Q S, et al.. Key wavelengths screening using competitive adaptive reweighted sampling method for multivariate calibration [J].Anal.Chim. Acta, 2009,648(1):77-84.
[29]	ZHU H Y, CHU B Q, ZHANG C,et al.. Hyperspectral imaging for presymptomatic detection of tobacco disease with successive projections algorithm and machine-learning classifiers [J/OL]. Sci. Rep., 2017, 7(1): 4125 [2024-05-06]. .
[30]	SAMPAIO P S, CASTANHO A, ALMEIDA A S, et al.. Identification of rice flour types with near-infrared spectroscopy associated with PLS-DA and SVM methods [J].Eur. Food Res.Technol., 2020, 246(3): 527-537.
[31]	GARCÍA-VERA Y E, POLOCHÈ-ARANGO A, MENDIVELSO-FAJARDO C A,et al.. Hyperspectral image analysis and machine learning techniques for crop disease detection and identification:a review [J/OL].Sustainability, 2024, 16(14): 6064 [2024-05-06]. .

处理 Trentment	训练集样本数 Number of training set samples	测试集样本数 Number of test set samples
水Water	149	35
高效氯氟氰菊酯Lambda-cyhalothrin	143	56
吡虫啉Imidacloprid	146	57
三唑酮Triadimefon	141	50
多菌灵Carbendazim	150	56
甲基硫菌灵Thiophanate-methyl	171	27

处理 Trentment	训练集样本数 Number of training set samples	测试集样本数 Number of test set samples
水Water	149	35
高效氯氟氰菊酯Lambda-cyhalothrin	143	56
吡虫啉Imidacloprid	146	57
三唑酮Triadimefon	141	50
多菌灵Carbendazim	150	56
甲基硫菌灵Thiophanate-methyl	171	27

预处理方法 Preprocessing method	训练集准确率 Training set accuracy/%	测试集准确率 Test set accuracy/%
原始数据Raw data	92.32	89.50
一阶导数D1	91.66	91.28
二阶导数D2	92.23	91.35
标准正态变换SNV	99.67	98.58
多重散射校正MSC	96.40	94.77

预处理方法 Preprocessing method	训练集准确率 Training set accuracy/%	测试集准确率 Test set accuracy/%
原始数据Raw data	92.32	89.50
一阶导数D1	91.66	91.28
二阶导数D2	92.23	91.35
标准正态变换SNV	99.67	98.58
多重散射校正MSC	96.40	94.77

预处理方法 Preprocessing method	训练集准确率 Training set accuracy/%	测试集准确率 Test set accuracy/%
原始数据Raw data	99.89	63.35
一阶导数D1	100.00	97.55
二阶导数D2	100.00	98.58
标准正态变换SNV	100.00	85.41
多重散射校正MSC	99.78	85.77