Please wait a minute...
浙江大学学报（农业与生命科学版）
浙江大学学报（农业与生命科学版）  2019, Vol. 45 Issue (2): 256-262    DOI: 10.3785/j.issn.1008-9209.2018.05.151
农业工程     
基于卷积神经网络模型的大豆花叶病初期高光谱检测
桂江生1,2(),吴子娴1,李凯2
1. 浙江理工大学信息学院，杭州 310018
2. 南京农业大学农学院，南京 210095
Hyperspectral imaging for early detection of soybean mosaic disease based on convolutional neural network model
Jiangsheng GUI1,2(),Zixian WU1,Kai LI2
1. School of Information Science and Technology, Zhejiang Sci-Tech University, Hangzhou 310018, China
2. College of Agriculture, Nanjing Agricultural University, Nanjing 210095, China
 全文: PDF(2732 KB)   HTML ( )  HTML
摘要：

为减轻花叶病对大豆产量的影响，实现对大豆花叶病害初期的快速检测，本文提出了一种基于卷积神经网络（convolutional neural network, CNN）模型的大豆花叶病害的诊断识别方法。首先对分别接种SC3、SC7病毒7 d后发病初期及正常的‘南农1138-2’大豆样本各80片（共计240片）进行高光谱图像采集，根据其图像信息提取并计算感兴趣区域的平均光谱值，建立基于高光谱图像的CNN模型。最终模型训练集识别率达到94.79%，预测集识别率达到92.08%，其中对接种SC3病毒的花叶病叶片识别率为88.75%，对接种SC7病毒的花叶病叶片识别率为93.13%，对正常叶片识别率为94.38%。对比最小二乘支持向量机和极限学习机模型，CNN模型能够更充分提取光谱的深层特征信息，识别效果显著提高。研究表明，基于高光谱图像的CNN模型能够更精确地实现对大豆花叶病初期检测，将CNN与高光谱结合的方法也为病害检测提供了一种新思路。
关键词： 大豆;  花叶病;  高光谱检测;  卷积神经网络    
Abstract:

In order to reduce the impact of mosaic disease on soybean production and explore a theoretical basis for rapid detection of early soybean mosaic disease, a novel hyperspectral detection method for early soybean mosaic disease based on convolutional neural network (CNN) model was proposed. First, soybean samples inoculated separately with SC3, SC7 viruses and normal soybean samples (Nannong 1138-2) were collected through a hyperspectral system. A region of 40 pixel×40 pixel was selected as the region of interest (ROI) and the average spectral information of ROI was extracted. Then, the CNN model was established based the hyperspectral image. Finally, the recognition rate of the training set in the CNN model reached 94.79%, and the recognition rate of the prediction set reached 92.08%. The recognition rate of the mosaic leaf inoculated with SC3 virus was 88.75%, and the recognition rate of the mosaic leaf inoculated with SC7 virus was 93.13%, and the recognition rate of the normal leaf was 94.38%. Compared with the least square-support vector machine (LS-SVM) and extreme learning machine (ELM) models, the CNN model can more fully extract the deep features of the spectrum, and the extracting effect was significantly improved. Thus, this research shows that the CNN model can achieve the detection of early soybean mosaic disease more accurately, and combining the CNN model with hyperspectral methods also provides a new idea for plant disease detection.
Key words: soybean    mosaic disease    hyperspectral detection    convolutional neural network
收稿日期: 2018-05-15 出版日期: 2019-04-25
CLC:  S 431.9  
基金资助: 国家自然科学基金(61105035，61502430)
通讯作者: 桂江生     E-mail: jsgui@zstu.edu.cn
作者简介: 桂江生（https://orcid.org/0000-0003-2455-3390），E-mail： jsgui@zstu.edu.cn
服务　　
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章　　
桂江生
吴子娴
李凯

引用本文:

桂江生,吴子娴,李凯. 基于卷积神经网络模型的大豆花叶病初期高光谱检测[J]. 浙江大学学报（农业与生命科学版）, 2019, 45(2): 256-262.

Jiangsheng GUI,Zixian WU,Kai LI. Hyperspectral imaging for early detection of soybean mosaic disease based on convolutional neural network model. Journal of Zhejiang University (Agriculture and Life Sciences), 2019, 45(2): 256-262.

链接本文:

http://www.zjujournals.com/agr/CN/10.3785/j.issn.1008-9209.2018.05.151    或    http://www.zjujournals.com/agr/CN/Y2019/V45/I2/256

图1  高光谱成像系统
图2   LeNet卷积网络的结构
图3  卷积神经网络处理流程
图4  大豆样本高光谱图像
图5   3种大豆样本的平均光谱曲线

模型

model

 识别率 Recognition rate

训练集

Training set

预测集

Prediction set

接种SC3叶

Leaf inoculated with SC3

接种SC7叶

Leaf inoculated with SC7

正常叶

Normal leaf
LS-SVM 86.70 85.63 82.50 86.25 88.13
ELM 89.59 87.50 85.00 88.13 89.38
CNN 94.79 92.08 88.75 93.13 94.38
表1  不同分类模型的检测结果
1 智海剑,盖钧镒 .大豆花叶病毒及抗性遗传的研究进展.大豆科学,2006,25(2):174-180.
ZHI H J , GAI J Y . Research progress on genetics of soybean mosaic virus and resistance. Soybean Science, 2006,25(2):174-180. (in Chinese with English abstract)
2 CROSSLIN J M , LIN H , MUNYANEZA J E . Detection of ‘Candidatus Liberibacter solanacearum’ in the potato psyllid, Bactericera cockerelli (Sulc), by conventional and real-time PCR. Southwestern Entomologist, 2011,36(2):125-135.
3 MASSART S , OlMOS A , JIJAKLI H , et al . Current impact and future directions of high throughput sequencing in plant virus diagnostics. Virus Research, 2014,188:90-96.
4 王文桥,马平,张小风,等 .生物源杀菌剂与化学药剂协调防控番茄病害.植物保护学报,2011,38(1):75-80.
WANG W Q , MA P, ZHANG X F , et al . Biological source fungicides and chemical agents coordinated to prevent tomato diseases. Journal of Plant Protection, 2011,38(1):75-80. (in Chinese with English abstract)
5 RAVICHANDRAN N K , WIJESINGHE R E , SHIRAZI M F , et al . In vivo monitoring on growth and spread of gray leaf spot disease in Capsicum annuum leaf using spectral domain optical coherence tomography. Journal of Spectroscopy, 2016(4):1-6.
6 GARCIA-RUIZ F , SANKARAN S , MAJA J M , et al . Comparison of two aerial imaging platforms for identification of Huanglongbing-infected citrus trees. Computers & Electronics in Agriculture, 2013,91:106-115.
7 王献锋,张善文,王震,等 .基于叶片图像和环境信息的黄瓜病害识别方法.农业工程学报,2014,30(14):148-153.
WANG X F , ZHANG S W , WANG Z , et al . Recognition method of cucumber diseases based on leaf image and environmental information. Transactions of the CSAE, 2014,30(14):148-153. (in Chinese with English abstract)
8 ADEBAYO S E , HASHIM N , ABDAN K , et al . Application and potential of backscattering imaging techniques in agricultural and food processing:a review. Journal of Food Engineering, 2016,169:155-164.
9 谢传奇,方孝荣,邵咏妮,等 .番茄叶片早疫病近红外高光谱成像检测技术.农业机械学报,2015,46(3):315-319.
XIE C Q , FANG X R , SHAO Y N , et al . Detection of early blight on tomato leaves using near-infrared hyperspectral imaging technique. Journal of Agricultural Machinery, 2015,46(3):315-319. (in Chinese with English abstract)
10 KRISHNA G , SAHOO R N , PARGAL S , et al . Assessing wheat yellow rust disease through hyperspectral remote sensing. ISPRS Technical CommissionⅧSymposium at Hyderabad, Pakistan, 2014. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, 2014,XL- 8:1413-1416.
11 CAPORASO N , WHITWORTH M B , GREBBY S , et al . Rapid prediction of single green coffee bean moisture and lipid content by hyperspectral imaging. Journal of Food Engineering, 2018,227:18-29.
12 ZHANG J , WANG N , YUAN L , et al . Discrimination of winter wheat disease and insect stresses using continuous wavelet features extracted from foliar spectral measurements. Biosystems Engineering, 2017,162:20-29.
13 LIAGHAT S , EHSANI R , MANSOR S , et al . Early detection of basal stem rot disease (Ganoderma) in oil palms based on hyperspectral reflectance data using pattern recognition algorithms. International Journal of Remote Sensing, 2014,35(10):3427-3439.
14 刘思伽,田有文,张芳,等 .采用二次连续投影法和BP人工神经网络的寒富苹果病害高光谱图像无损检测.食品科学,2017,38(8):277-282.
LIU S J , TIAN Y W , ZHANG F , et al . Hyperspectral imaging for nondestructive detection of Hanfu apple diseases using successive projections algorithm and BP neural network. Food Science, 2017,38(8):277-282. (in Chinese with English abstract)
15 ZHAO Y R , LI X , YU K Q , et al . Hyperspectral imaging for determining pigment contents in cucumber leaves in response to angular leaf spot disease. Scientific Reports, 2016,6(1):27790-27799.
16 CAO X , ZHOU F , XU L , et al . Hyperspectral image classification with Markov random fields and a convolutional neural network. IEEE Transactions on Image Processing, 2018,27(5):2354-2367.
17 CHEN Y , JIANG H , LI C , et al . Deep feature extraction and classification of hyperspectral images based on convolutional neural networks. IEEE Transactions on Geoscience & Remote Sensing, 2016,54(10):6232-6251.
18 YU S Q , JIA S , XU C Y . Convolutional neural networks for hyperspectral image classification. Neurocomputing, 2017,219:88-98.
19 SAVITZKY A , GOLAY M J E . Smoothing and differentiation of data by simplified least squares procedures. Analytical Chemistry, 1964,36(8):1627-1639.
20 CHATFIELD K , SIMONYAN K , VEDALDI A , et al . Return of the devil in the details: delving deep into convolutional nets. Eprint Arxiv, 2014:1-11.
21 VOLPI M , TUIA D . Dense semantic labeling of subdecimeter resolution images with convolutional neural networks. IEEE Transactions on Geoscience & Remote Sensing, 2016(10):1-13.
22 KRIZHEVSKY A , SUTSKEVER I , HINTON G E . ImageNet classification with deep convolutional neural networks: NIPS’12 Proceedings of the 25th International Conference on Neural Information Processing Systems, Nevada, US, December 3-6, 2012. [S. l.]: [s. n.] 2012:1097-1105.
23 MAO Q , DONG M , HUANG Z , et al . Learning salient features for speech emotion recognition using convolutional neural networks. IEEE Transactions on Multimedia, 2014,16(8):2203-2213.
24 JADERBERG M , SIMONYAN K , VEDALDI A , et al . Reading text in the wild with convolutional neural networks. International Journal of Computer Vision, 2014,116(1):1-20.
25 SUYKENS J A K , GESTEL T V , BRABANTER J D , et al . Least square support vector machine. Euphytica, 2002,2(2):1599-1604.
26 HUANG G B , ZHU Q Y , SIEW C K . Extreme learning machine: theory and applications. Neurocomputing, 2006,70(1/2/3):489-501.
27 杨丹,赵海滨,龙哲 .MATLAB图像处理实例详解.北京:清华大学出版社,2013.
YANG D , ZHAO H B , LONG Z . Detailed Explanation of MATLAB Image Processing. Beijing: Tsinghua University Press, 2013. (in Chinese)
28 沈焕锋 .ENVI遥感影像处理方法.武汉:武汉大学出版社,2009.
SHEN H F . ENVI Remote Sensing Image Processing Method. Wuhan: Wuhan University Press, 2009. (in Chinese)
29 黄崑 .Excel统计分析基础教程.北京:清华大学出版社,2011.
HUANG K . Excel Statistical Analysis Basic Tutorial. Beijing: Tsinghua University Press, 2011. (in Chinese)
[1] 庞婷，帅鹏，陈平，杜青，付智丹，杨文钰，雍太文. 不同结瘤品种和行间距对套作大豆根瘤生长及物质积累与分配的影响[J]. 浙江大学学报(农业与生命科学版), 2017, 43(4): 451-461.
[2] 方萍,刘卫国,刘孝德,刘婷,池晓玉,许燕,庞婷,彭霄,蔡凌,杨文钰. 玉-豆间作对菜用大豆品质的影响[J]. 浙江大学学报(农业与生命科学版), 2016, 42(5): 556-.
[3] 程云清1*, 张奇2， 刘剑锋1， 张会弟1， 张春吉1. 外源乙烯调控大豆花粉育性的研究[J]. 浙江大学学报(农业与生命科学版), 2014, 40(1): 25-32.
[4] 唐秀莹1， 陈正礼1，2，3*， 罗启慧2，3， 张小龙1. 大豆异黄酮对大鼠肠道上皮内淋巴细胞、杯状细胞及瘦素长型受体的影响[J]. 浙江大学学报(农业与生命科学版), 2013, 39(3): 343-350.
[5] 马林1， 周练1， 周正剑1， 唐桂香2， 沈志成2， 寿惠霞1.   抗除草剂转基因水稻和大豆快速准确检测技术研究[J]. 浙江大学学报(农业与生命科学版), 2012, 38(6): 647-654.
[6] 范觉鑫1,2， 张彬2， 李丽立1， 袁晓雪1， 耿梅梅1， 罗佳捷2. 大豆异黄酮对雄性香猪生殖器官发育及组织生化指标的影响[J]. 浙江大学学报(农业与生命科学版), 2012, 38(4): 477-484.
[7] 雷婷， 向达兵， 郭凯， 杨文钰， 刘增禹， 陈小容. 磷钾对套作大豆干物质积累与分配及产量的影响[J]. 浙江大学学报(农业与生命科学版), 2012, 38(3): 318-328.
[8] 万燕,闫艳红,杨文钰. 不同氮肥水平下叶面喷施烯效唑对套作大豆生长和氮代谢的影响[J]. 浙江大学学报(农业与生命科学版), 2012, 38(2): 185-196.
[9] 尚海丽,周雪平,吴建祥;. 免疫斑点法和免疫捕获RT-PCR检测黄瓜绿斑驳花叶病毒[J]. 浙江大学学报(农业与生命科学版), 2010, 36(5): 485-490.
[10] 李彬,粟寒,吴翠萍,周明华,安榆林;. 一种豇豆重花叶病毒IC-RT real-time PCR检测方法的建立[J]. 浙江大学学报(农业与生命科学版), 2010, 36(5): 491-496.
[11] 程云清,杨 君,安利佳,陈智文,刘剑锋. 乙烯对大豆产量及花器官发育的影响[J]. 浙江大学学报(农业与生命科学版), 2008, 34(5): 540-545.
[12] 金剑　王光华　刘晓冰等. 1950-2006年间黑龙江省大豆品种农艺性状的演变[J]. 浙江大学学报(农业与生命科学版), 2008, 34(3): 296-302.
[13] 施曼玲　 周雪平. 芜菁花叶病毒两分离物的CP基因及HC-Pro基因的比较[J]. 浙江大学学报(农业与生命科学版), 2005, 31(6): 689-653.
[14] 刘鹏　 吴建之　 杨玉爱. 钼、硼供给水平对大豆钼、硼吸收与分配的影响[J]. 浙江大学学报(农业与生命科学版), 2005, 31(4): 399-407.
[15] 韩立德　 胡晋　 徐海明　 邱家驯　 黄正来. 夏播菜用大豆感官品质性状核心种质构建方法的研究[J]. 浙江大学学报(农业与生命科学版), 2005, 31(3): 288-292.