| 299 | 0 | 69 |
| 下载次数 | 被引频次 | 阅读次数 |
为降低烟叶霉变对卷烟产品品质的影响，建立了一种烟叶霉变无损检测方法。使用高光谱成像仪采集受霉变影响的烟叶高光谱图像并提取光谱数据，采用归一化（Min-Max Scaler,MMS）、标准正态变化（Standard Normal Variate,SNV）、多元散射校正（Multivariate Scattering Correction,MSC）和平滑滤波（Savitzky-Golay,SG）等7种方法，对获得的烟叶光谱数据进行预处理；利用连续投影变换（Successive Projections Algorithm,SPA）、主成分分析（Principal Component Analysis,PCA）等算法对特征波长进行选择，并运用随机森林（Random Forest,RF）、支持向量机（Support Vector Machine,SVM）等机器学习算法建立6种分类模型。将基于7种预处理方法建立的6种分类模型进行对比测试，结果表明：(1)SNV为最优光谱预处理方法，而基于SPA选择特征波长建立的RF模型性能优异，SPA-RF模型在训练集和测试集上的识别准确率分别达到98.82%和98.64%，降低了基于全波长分类模型的运算时间，对于相同等级不同产地烟叶具有良好分类结果；(2)高光谱成像技术结合SPA-RF模型可有效实现烟叶霉变情况的准确识别。该技术可为提高烟叶品质管控质量提供支持。
Abstract：To mitigate the impact of moldy tobacco leaves on the quality of cigarette products, a non-destructive method for detecting moldy tobacco leaves was established. Hyperspectral images of moldy tobacco leaves were acquired using a hyperspectral imager, and the spectral data were extracted.The acquired spectral data were pre-processed with seven methods including Min-Max Scaling(MMS),Standard Normal Variate(SNV), Multivariate Scatter Correction(MSC), and Savitzky-Golay smoothing filter(SG). The characteristic wavelengths were selected with Successive Projections Algorithm(SPA) and Principal Component Analysis(PCA). Six classification models were established using machine learning algorithms including Random Forest(RF) and Support Vector Machine(SVM), and comparatively tested.The results showed that: 1) SNV was the optimal spectral preprocessing method, and the RF model based on characteristic wavelengths selected by SPA exhibited superior performance, achieved recognition accuracies of 98.82% and 98.64% on the training and test sets, respectively. This approach also reduced the computational time of full-spectrum classification algorithms and achieved favorable classification results for tobacco of the same grade from different growing areas. 2) The hyperspectral imaging technology combined with the SPA-RF model could effectively and accurately identify the moldy states of tobacco leaves. This technology supports the promotion of tobacco leaf quality control and management.
[1]陈二龙，范志勇，宋朝鹏，等.烘烤期烟叶霉变发生规律及关键影响因素[J].江西农业大学学报，2017,39(5):877-883.CHEN Erlong,FAN Zhiyong,SONG Zhaopeng,et al.An analysis of the law and key influencing factors of tobacco leaf mildew during baking[J]. Acta Agriculturae Universitatis Jiangxiensis,2017,39(5):877-883.
[2]孔凡玉，林建胜，张成省，等.储烟霉变机理与防霉技术研究进展[J].中国烟草学报，2009,15(5):78-81.KONG Fanyu,LIN Jiansheng,ZHANG Chengsheng,et al. Recent progress in research on mechanism and control of tobacco mildew during storage[J]. Acta Tabacaria Sinica,2009,15(5):78-81.
[3]黄阔，朱贝贝，叶长文，等.雪茄烟叶工业发酵过程中霉变对真菌群落结构的影响研究[J].轻工学报，2024,39(1):73-78,96.HUANG Kuo, ZHU Beibei, YE Changwen, et al.The effect of mold on the fungal community structure during the industrial fermentation of cigar tobacco leaves[J]. Journal of Light Industry,2024,39(1):73-78,96.
[4]翟乃琦，云利军，叶志霞，等.基于一维卷积神经网络的烟叶仓储霉变预测方法研究[J].计算机工程与科学，2021,43(10):1833-1837.ZHAI Naiqi,YUN Lijun,YE Zhixia,et al. A tobacco storage moldy prediction method based on one-dimensional convolutional neural network[J].Computer Engineering and Science, 2021, 43(10):1833-1837.
[5]赖燕华，林云，陶红，等.烟叶霉变的快速识别——基于近红外光谱与随机森林算法[J].中国烟草学报，2020,26(2):36-43.LAI Yanhua, LIN Yun, TAO Hong, et al. Rapid identification of tobacco mildew based on near infrared spectroscopy and random forest algorithm[J]. Acta Tabacaria Sinica,2020,26(2):36-43.
[6]周继月，杨盼盼，刘磊，等.基于近红外光谱分析技术结合化学计量学方法的初烤烟叶霉变预测研究[J].中国烟草学报，2018,24(1):6-13.ZHOU Jiyue,YANG Panpan,LIU Lei,et al. Prediction model for flue-cured tobacco leaf mildew based on NIR spectroscopy combined with chemometrics[J]. Acta Tabacaria Sinica,2018,24(1):6-13.
[7]吴永清，李明，张波，等.高光谱成像技术在谷物品质检测中的应用进展[J].中国粮油学报，2021,36(5):165-173.WU Yongqing, LI Ming, ZHANG Bo, et al.Application of hyperspectral imaging technology in grain quality detection[J]. Journal of the Chinese Cereals and Oils Association,2021,36(5):165-173.
[8]李跑，谭惠珍，谢叔娥，等.基于近红外光谱技术有监督模式识别的青皮产地溯源分析[J].轻工学报，2024,39(2):54-59.LI Pao,TAN Huizhen,XIE Shue,et al. Traceability analysis of Pericarpium Citri Reticulatae Viride origin based on near infrared spectroscopy technology and supervised pattern recognition[J]. Journal of Light Industry,2024,39(2):54-59.
[9]李斌，苏成涛，殷海，等.高光谱成像技术结合机器学习的稻米霉变检测[J].光谱学与光谱分析，2023,43(8):2391-2396.LI Bin,SU Chengtao,YIN Hai,et al. Hyperspectral imaging technology combined with machine learning for detection of moldy rice[J]. Spectroscopy and Spectral Analysis,2023,43(8):2391-2396.
[10]金诚谦，郭榛，张静，等.大豆水分含量的高光谱无损检测及可视化研究[J].光谱学与光谱分析，2022,42(10):3052-3057.JIN Chengqian, GUO Zhen, ZHANG Jing, et al.Non-destructive detection and visualization of soybean moisture content using hyperspectral technique[J].Spectroscopy and Spectral Analysis, 2022, 42(10):3052-3057.
[11]刘燕德，王舜.基于图像和光谱融合的脐橙货架期高光谱成像无损检测研究[J].光谱学与光谱分析，2022,42(6):1792-1797.LIU Yande,WANG Shun. Research on non-destructive testing of navel orange shelf life imaging based on hyperspectral image and spectrum fusion[J].Spectroscopy and Spectral Analysis, 2022, 42(6):1792-1797.
[12]张悦，李扬，宋月鹏.苹果轻微机械损伤高光谱无损检测Stacking集成学习模型构建[J].光谱学与光谱分析，2023,43(7):2272-2277.ZHANG Yue,LI Yang,SONG Yuepeng. Nondestructive detection of slight mechanical damage of apple by hyperspectral spectroscopy based on stacking model[J].Spectroscopy and Spectral Analysis, 2023, 43(7):2272-2277.
[13] Yang D,Yuan J H,Chang Q,et al. Early determination of mildew status in storage maize kernels using hyperspectral imaging combined with the stacked sparse auto-encoder algorithm[J]. Infrared Physics&Technology,2020,109:103412.
[14] Zhao J L,Fang Y,Chu G M,et al. Identification of leaf-scale wheat powdery mildew(Blumeria graminis f.sp. Tritici)combining hyperspectral imaging and an SVM classifier[J]. Plants,2020,9(8):936.
[15] Wei X, Deng C J, Fang W, et al. Classification method for folded flue-cured tobacco based on hyperspectral imaging and conventional neural networks[J]. Industrial Crops and Products,2024,212:118279.
[16] 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]. Processes,2023,11(4):1249.
[17] Liang J,Wang Y Y,Shi Y,et al. Non-destructive discrimination of homochromatic foreign materials in cut tobacco based on VIS-NIR hyperspectral imaging[J]. Journal of the Science of Food and Agriculture,2023,103(9):4545-4552.
[18] Marcelo A C M,Soares F L F,Ardila A J,et al. Fast inline tobacco classification by near-infrared hyperspectral imaging and support vector machine-discriminant analysis[J]. Analytical Methods,2019,11(14):1966-1975.
[19] Chen H T,Han Y J,Liu Y C,et al. Classification models for tobacco mosaic virus and potato virus Y using hyperspectral and machine learning techniques[J]. Frontiers in Plant Science,2023,14:1211617.
[20]李鑫，汤卫荣，张永辉，等.基于高光谱成像技术的烟叶田间成熟度判别模型[J].烟草科技，2022,55(7):17-24.LI Xin, TANG Weirong, ZHANG Yonghui, et al.Discriminant model for field maturity of tobacco leaves based on hyperspectral imaging technology[J]. Tobacco Science&Technology,2022,55(7):17-24.
[21]李士静，潘羲，陈熙卓，等.基于高光谱信息的烟叶分级方法比较[J].烟草科技，2021,54(10):82-91.LI Shijing,PAN Xi,CHEN Xizhuo,et al. Comparison of tobacco grading methods based on hyperspectral information[J]. Tobacco Science&Technology,2021,54(10):82-91.
[22]梅吉帆，郭文孟，李智慧，等.基于高光谱成像的烤烟着生部位识别[J].中国烟草学报，2024,30(3):51-60.MEI Jifan,GUO Wenmeng,LI Zhihui,et al. Growth position discrimination of flue-cured tobacco leaves based on hyperspectral imaging[J]. Acta Tabacaria Sinica,2024,30(3):51-60.
[23]郭文孟，薛宇毅，罗靖，等.基于高光谱成像的烟叶泛青特征分析与表征[J].烟草科技，2023,56(7):84-91.GUO Wenmeng,XUE Yuyi,LUO Jing,et al. Analysis and characterization of tobacco leaf greenish based on hyperspectral imaging[J]. Tobacco Science&Technology,2023,56(7):84-91.
[24]梅吉帆，李智慧，李嘉康，等.基于高光谱成像技术的配方烟丝组分判别[J].分析测试学报，2021,40(8):1151-1157.MEI Jifan,LI Zhihui,LI Jiakang,et al. Components discrimination for formula tobacco based on hyperspectral imaging[J]. Journal of Instrumental Analysis,2021,40(8):1151-1157.
[25]李智慧，梅吉帆，李辉，等.高光谱成像的非烟物质分类识别研究[J].中国烟草学报，2022,28(3):81-88.LI Zhihui, MEI Jifan, LI Hui, et al. Research on classification and recognition of non-tobacco related material(NTRM)based on hyperspectral imaging technology[J]. Acta Tabacaria Sinica, 2022, 28(3):81-88.
[26] Abeywickrama T,Cheema M A,Taniar D. K-nearest neighbors on road networks:A journey in experimentation and in-memory implementation[J].Proceedings of the VLDB Endowment,2016,9(6):492-503.
[27] Tao D L,Xiao Z T,Zhang F F,et al. Cloth defect classification method based on SVM[J]. International Journal of Digital Content Technology and Its Applications,2013,7(3):614-622.
[28] Billari F C. A log-logistic regression model for a transition rate with a starting threshold[J]. Population Studies,2001,55(1):15-24.
[29]杜瑞杰，王双成，高瑞.基于高斯密度的一阶贝叶斯衍生分类器[J].计算机应用研究，2015,32(11):3242-3246.DU Ruijie,WANG Shuangcheng,GAO Rui. First-order Bayesian derivative classifier based on Gaussian density[J]. Application Research of Computers,2015,32(11):3242-3246.
[30] Bazan J G,Bazan-Socha S,Buregwa-Czuma S,et al. A classifier based on a decision tree with verifying cuts[J].Fundamenta Informaticae,2016,143(1/2):1-18.
[31] Breiman L. Random forests[J]. Machine Learning,2001,45(1):5-32.
基本信息:
DOI：10.16135/j.issn1002-0861.2024.0317
中图分类号:TS47;TP391.41;TP181
引用信息:
[1]范鹏飞,马建伟,姚思愚等.基于高光谱成像和机器学习的烟叶霉变检测方法[J].烟草科技,2024,57(12):96-105.DOI:10.16135/j.issn1002-0861.2024.0317.
基金信息:
河南省科技攻关计划“基于深度学习和高光谱成像的烟叶原料霉变无损检测关键技术研究及应用”（242102220028）; 中国烟草总公司重大科技项目“面向工业需求和智能分选的烟叶分级大数据研究与构建”（110202201052）