| 369 | 3 | 46 |
| 下载次数 | 被引频次 | 阅读次数 |
【目的】采用高光谱成像技术结合机器学习方法，建立烤烟着生部位（上部、中部、下部）的识别模型。【方法】首先，通过分析烟叶在水、氮敏感波段下的强度分布特征，采用了一种结合OTSU和Sauvola图像分割算法的双阈值感兴趣区（ROI）选取方法，然后对比分析不同预处理方法对数据建模的影响规律，采用支持向量机（SVM）、极限梯度提升（XGBoost）算法进行判别模型的建立，通过参数寻优进行模型的优化。使用遗传算法（GA）和遗传算法结合连续投影算法（GA-SPA）进行特征波长的选择，建立简化模型。【结果】（1）建立的双阈值感兴趣区选取方法能准确高效地实现烤烟叶片正常叶面区域的选取（2）不同数据预处理方法对识别模型影响较为显著，基于一阶导和萨维莱茨-戈莱平滑（1Der+SG）预处理光谱数据，结合GA选取的特征波长建立的XGBoost着生部位识别模型具有最佳的分类效能，其准确率高达97.78%。【结论】研究建立的基于高光谱成像技术结合机器学习方法的部位模型可满足烤烟着生部位的高效准确识别。
Abstract：[Background] Using hyperspectral imaging technology combined with machine learning methods, a recognition model for the growth potion(upper, middle, lower) of flue-cured tobacco was established. [Methods] The intensity distribution characteristics of tobacco leaves in water and protein sensitive bands was analyzed firstly. Then, a two-threshold region of Interest(ROI) selection method combining OTSU and Sauvola image segmentation algorithms was raised based on the analysis of abundance distribution. Moreover, the influences of different data preprocessing methods on data modeling were comparatively analyzed. Support vector machine(SVM) and extreme gradient boosting(XGBoost) algorithm were adopted to establish the discrimination model, and the models were optimized by parameter optimization subsequently. Using the Genetic Algorithm(GA) and the combination of Genetic Algorithm with Successive Projection Algorithm(GA-SPA) for the selection of characteristic wavelengths, a simplified model was established. [Results] Test data show: 1) The established dual threshold method can efficiently select the right leaf area of flue-cured tobacco leaves; 2) the efficacy of discrimination model was obviously affected by data preprocessing methods. Based on the spectral data preprocessed using the first-order derivative and Savitzky-Golay smoothing(1Der+SG), combined with the characteristic wavelengths selected by GA, the XGBoost recognition model for the growth position had the best classification performance, with an accuracy rate of up to 97.78%. [Conclusion] The model established in this study, based on hyperspectral imaging technology combined with machine learning methods, can efficiently and accurately identify the growth position of flue-cured tobacco.
[1]MAKKAR H,BECKER K.Nutrients and antiequality factors in different morphological parts of the Moringa oleifera tree[J].The Journal of Agricultural Science,1997,128(3):311-322.
[2]孟哲，王丽君，黄洋，等.中药三七不同药用部位的指纹图谱及模式识别研究[J].中国药学杂志，2020,55(07):504-509.MENG Zhe,WANG Li Jun,HUANG Yang,et al.Analysis of fingerprint and pattern recognition of different parts of panax notoginseng[J].Chinese Pharmaceutical Journal,2020,55(07):504-509.
[3]吴佳，叶文卉，卢宗元，等.藤茶不同部位指纹图谱及其差异研究[J].天然产物研究与开发，2020,32(09):1507-1514WU Jia,YE Wenghui,LU Zongyuan,et al.Research on fingerprints and differences of different parts of Ampelopsis grossedentata[J].Natural Product Research And Development,2020,32(09):1507-1514.
[4]高敏霞，王小安，叶新福，等.不同着生部位辣木叶营养成分差异分析[J].南方农业学报，2017,48(08):1488-1492.GAO Mingxia,WANG Xiaoan,YE Xingfu,et al.Variance analysis for nutritional components in Moringa oleifera Lam.leaves picked from different positions of branches[J].Journal of Southern Agriculture,2017,48(08):1488-1492.
[5]沙云菲，赵亚萍，于洁，等.基于支持向量机的烟叶不同部位6种化学成分差异性分析[J].东华大学学报：自然科学版，2019,45(05):720-723.SHA Yunfei,ZHAO Yapin,YU Jie,et al.Difference analysis of six chemical composition in tobacco parts based on support vector machine[J].Journal of Donghua University (Natural Science),2019,45(05):720-723.
[6]张鑫，郭佳，倪力军，等.基于红外与近红外光谱的烟叶部位识别[J].光谱学与光谱分析，2007,21(12):2437-2440.ZHANG Xin,GUO Jia,NI Lijun,et al.Recognition of plant parts of tobaccos based on infrared and near infrared spectra[J].Spectroscopy and spectral analysis,2007,27(12):2437-2440.
[7]阮春生，张强，丁波洋，等.不同部位烟叶紫外-可见光谱特征的研究[J].云南民族大学学报：自然科学版，2009,18(03):237-241.RUAN Chunsheng,ZHANG Qiang,DING Boyang,et al.A study of the UV-VIS spectral features of different tobacco leaf[J].Journal of Yunnan Nationalities University:Natural Science,2009,18(03):237-241.
[8]余春霞，马翔，张晔晖，等.基于近红外光谱和SIMCA算法的烟叶部位相似性分析[J].光谱学与光谱分析，2011,31(04):924-927.YU Chunxia,MA Xiang,ZHANG Yehui,et al.Tobacco plant parts similarity analysis based on near-infrared spectroscopy and SIMCAalgorithm[J].Spectroscopy and spectral analysis,2011,31(04):924-927
[9]蔡嘉月，梁淼，温亚东，等.应用可见-近红外高光谱分析烟叶的颜色和部位特征[J].光谱学与光谱分析，2014,34(10):2758-2763.CAI Jiayue,LIANG Miao,WEN Yadong,et al.Analysis of tobacco color and location features using visible-near infrared hyperspectral data[J].Spectroscopy and spectral analysis,2014,34(10):2758-2763.
[10]董惠忠，毕艳玖，赵晓华，等.基于致香成分的上部烟叶和中部烟叶分类判别[J].上海大学学报：自然科学版，2019,25(02):309-316.DONG Huizhong,BI Yanjiu,ZHAO Xiaohua,et al.Classification of upper and middle leaves of tobaccos based on aromatic components[J].Journal of Shanghai University:Natural Science Edition,2019,25(02):309-316.
[11]刘立新，何迪，李梦珠，等.基于高光谱技术与机器学习的新疆红枣品种鉴别[J].中国激光，2020,47(11):291-298.LIU Lixin,HE Di,LI Mengzhu,et al.Identification of Xinjiang Jujube varieties based on hyperspectral technique and machine learning[J].Chinese Journal of Lasers,2020,47(11):291-298.
[12]牛文娟.基于图像处理的烟叶分级研究[D].郑州大学,2010.NIU Wengjuan.Tobacco classification based on image processing[D].Zhengzhou University,2010.
[13]郭强.基于图像处理和神经网络的烟叶分级研究[D].中南大学，2013GUO Qiang.The research of tobacco grading based on image processing and neural network[D].Central South University,2013.
[14]苏明秋.基于烟叶图像的参数准确提取和分级识别系统研究与实现[D].电子科技大学，2020.SU Mingqiu.Research and implementation of parameter extraction and classification recognition system based on tobacco image[D].University of electronic science and technology of China,2020.
[15]孙大文，吴迪，何鸿举，等.现代光学成像技术在食品品质快速检测中的应用[J].华南理工大学学报：自然科学版，2012,40(10):59-68SUN Dawen,WU di,HE Hongju,et al.Applications of modern optical imaging technologies for rapid quality inspection of food products.Journal of South China University of Technology:Natural Science Edition,2012,40(10):59-68.
[16]Kim M S,Chen Y R,Mehl P M.Hyperspectral Reflectance and Fluorescence Imaging System for Food Quality and Safety[J].Transactions of the ASAE.American Society of Agricultural Engineers,2001,44(3):721-729.
[17]Kim M S,Cho B K,Lefcourt A M,et al.Multispectral fluorescence lifetime imaging of feces-contaminated apples by time-resolved laser-induced fluorescence imaging system with tunable excitation wavelengths[J].Applied Optics,2008,47(10):1608-1611.
[18]Barbin D F,Elmasry G,Sun D W,et al.Non-destructive determination of chemical composition in intact and minced pork using near-infrared hyperspectral imaging[J].Food Chemistry,2013,138(2-3):1162-1171.
[19]MA Ji,SUN D W.Prediction of monounsaturated and polyunsaturated fatty acids of various processed pork meats using improved hyperspectral imaging technique[J].Food Chemistry,2020,321,126695
[20]Pereira C S,Morais R,Reis M.Deep Learning Techniques for Grape Plant Species Identification in Natural Images[J].Sensors(Basel,Switzerland),2019,19(22):48-50
[21]李鸿强，孙红，李民赞.基于高光谱的马铃薯微型种薯分类检测[J].分析测试学报，2020,39(11):1421-1426.LI Hongqiang,SUN Hong,LI Mingzan.Classification detection of micro potato seed based on hyperspectral.Journal of Instrumental Analysis,2020,39(11):1421-1426.
[22]李梦竹.基于高光谱遥感的烤烟水分状况监测研究[D].河南农业大学，2017LI Mengzhu.Study on moisture monitoring of flue-cured tobacco based on hyperspectral remote sensing[D].Henan Agricultural University,2017.
[23]贾方方.不同光照处理对烤烟品质的影响及氮化物的高光谱监测研究[D].河南农业大学，2013JIA Fangfang.The effect of different light treatments on the quality of flue-cured tobacco and the monitoring of hyperspectrum on the nitrides.Henan Agricultural University,2013.
[24]MISHRA P,NORDON A,TSCHANNERL J,et al.Near-infrared hyperspectral imaging for non-destructive classification of commercial tea products[J].Journal of Food Engineering,2018,238(12):70-77.
[25]Rinnan S,Berg F,Engelsen S B.Review of the most common pre-processing techniques for near-infrared spectra[J].Tr ACTrends in Analytical Chemistry,2009,28(10):1201-1222.
[26]褚小立，袁洪福，王艳斌，等.遗传算法用于偏最小二乘方法建模中的变量筛选[J].分析化学，2001,29(04):437-442.ZHU Xiaoli,YUAN Hongfu,WANG Yanbin,et al.Variable selection for partial least squares modeling by genetic algorithm[J].Chinese Journal of Analytical Chemistry,2001,29(04):437-442.
[27]樊书祥，黄文倩，李江波，等.特征变量优选在苹果可溶性固形物近红外便携式检测中的应用[J].光谱学与光谱分析，2014,34(10):2707-2712.FAN Shuxiang,HUANG Wenqian,LI Jiangbo,et al.Application of characteristic NIR variables selection in portable detection of soluble solids contents of apple by near infrared spectroscopy[J].Spectroscopy and spectral analysis,2014,34(10):2707-2712.
[28]阳树洪.灰度图像阈值分割的自适应和快速算法研究[D].重庆大学，2014.YANG Shuhong.Study on the adaptive and fast algorithm of gray scale image thresholding.Chongqing University,2014.
[29]郭佳.基于光照不均匀图像的自适应二值化方法研究[D].武汉科技大学，2013.GUO Jia.The research of binarization method based on non-uniform illumination images.Wuhan University of Science and Technology,2013.
基本信息:
DOI：10.16472/j.chinatobacco.2022.033
中图分类号:S572;TP391.41;TP18
引用信息:
[1]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(03):51-60.DOI:10.16472/j.chinatobacco.2022.033.
基金信息:
福建中烟科技项目“卷烟产品及原料高光谱特征分析与应用技术研究”（No.D2020248）