基于高光谱成像技术的油菜种子油酸含量反演模型

doi:10.13304/j.nykjdb.2021.1001

摘要/Abstract

摘要：

高油酸油菜籽品种是当前油菜育种方向之一，为开发高效、无损测定油酸含量的方法，提高油菜高油酸种质资源筛选效率，选用3个油菜品种为材料，分别采集其种子光谱成像信息及油酸含量数据，首先对光谱信息进行11种预处理，确定多元散射校正（MSC）最佳预处理方法，然后基于主成分分析（PCA）、连续投影（SPA）、竞争性自适应重加权采样（CARS）方法对数据进行降维，最后分别建立支持向量机（SVM）、最小二乘支持向量机（LS-SVM）和极限学习机（ELM）3种定量分析模型，对油菜油酸含量进行无损检测。通过改变训练样本的数量来测试模型，为验证模型的稳定性，用相关系数（R）、均方根误差（RMSE）进行效果评价。结果表明，在所有模型中，多元散射校正+竞争性自适应重加权采样+极限学习机（MSC+CARS+ELM）模型预测效果最好，校正集相关系数（Rc）、均方根误差（RMSEc）分别为0.894、1.993 4%，预测集相关系数（Rp）为0.868，均方根误差（RMSEp）为1.069 8%，可更加准确地预测油酸含量，创建一种快速、无损检测油菜种子油酸含量的方法，为利用高光谱技术进行油菜营养品质无损检测提供理论依据。

关键词: 油菜, 高光谱成像, 油酸含量

Abstract:

Rape seed with high oleic acid breeding is one of the current breeding directions of rape. In order to find out an efficient and nondestructive method to analyze and determine the oleic acid content， and improve the screening efficiency of rape germplasm resources with high oleic acid， 3 rapeseed varieties were selected as materials in this study， and the spectral imaging information and oleic acid content data of seeds were collected， respectively. The spectral information was preprocessed in 11 ways， and multiplicative scatter correction CMSC was the optimal pretreatment method. 3 quantitative analysis models based on principal component analysis （PCA）， continuous projection （SPA）， competitive adaptive reweighted sampling （CARS） were established， and support vector machine （SVM）， and least square support vector machine （LSSVM） and extreme learning machine （ELM） were established to realize the nondestructive detection of rape oleic acid content. The model was tested by changing the number of training samples， and correlation coefficient （R） and root mean square error （RMSE） were used to evaluate the effect of these methods to validate the stability of the model. The results showed that MSC+CARS+ELM model had the best prediction effect among all the models. The correlation coefficient （Rc） and root mean square error RMSEc of correction set were 0.894 and 1.993 4%， respectively. The Rp and RMSEp of prediction set were 0.868 and 1.069 8%， respectively. It could more accurately predict the oleic acid content and provided theoretical basis for nondestructive testing of rapeseed nutritional quality by hyperspectral technology.

Key words: rape, hyperspectral imaging technology, oleic acid content

中图分类号:

S126

卢信, 廖桂平, 刘凡. 基于高光谱成像技术的油菜种子油酸含量反演模型[J]. 中国农业科技导报, 2023, 25(1): 92-99.

Xin LU, Guiping LIAO, Fan LIU. Inversion Model of Oleic Acid Content in Rape Seeds Based on Hyperspectral Imaging Technology[J]. Journal of Agricultural Science and Technology, 2023, 25(1): 92-99.

图/表 9

参考文献 26

1	李娜,杨涛.我国油菜籽产业发展现状及趋势展望[J].农业生产展望,2009,5 (2):15-17.
	LI N, YANG T. Rapeseed industry development present situation and trend in our country [J]. Agric. Production Outlook, 2009,5 (2):15-17.
2	张垚,李云昌,梅德圣,等.油菜油脂研究进展[J].植物学通报,2007,24(4):435-443.
	ZHANG Y, LI Y C, MEI D S, et al.. Advances in liplid research of rapesced [J]. Chin. Bull. Bot., 2007,24(4):435-443.
3	王东,王坤,吴静珠,等.基于光谱及成像技术的种子品质无损速测研究进展[J].光谱学与光谱分析,2021,41(1):52-59.
	WANG D, WANG K, WU J Z, et al.. Progress in nondestructive rapid measurement of seed quality based on spectroscopy and imaging technology [J]. Spect. Spectral. Anal., 2021,41(1):52-59.
4	徐扬帆,张秋平,徐明水,等.高含油高油酸油菜新品种‘帆鸣1号’选育及品质分析[J].分子植物育种,2020,18(15):5184-5190.
	XU Y F, ZHANG Q P, XU M S, et al.. The breeding and quality analysis of a new rape va iety fanming 1 with high oil conte nt and oleic acid content [J]. Mol. Plant Breeding, 2020,18(15):5184-5190.
5	杨柳,李东欣,谭太龙.不同油酸含量油菜脂肪酸含量分析[J].作物研究,2018,32(5):390-394, 402.
	YANG L, LI D X, TAN T L. Analysis of fat ty acid content in rapeseed with different oleic acid contents [Jl. Crop Res., 2018,32(5):390-394, 402.
6	曾静,伊雄海,曲栗,等.基于脂肪酸含量分析结合化学计量学的橄榄油等级鉴别方法[J].分析测试学报,2021,40(10):1432-1438.
	ZENG J, YI X H, QU L, et al.. Identification of olive oil grade based on fatty acid content analysis combined with stoichiometry [J]. J. Anal. Test., 2021,40(10):1432-1438.
7	李施蒙,王耐红,张敏.等 .甘蓝型油菜脂肪酸组分近红外分析模型构建[J].广东农业科学,2014(21):22-27.
	LI S M, WANG N H, ZHANG M, et al.. Establishment of near infrared analytic model formajor fatty acids content of Brassica napus L. [J]. Guangdong Agric. Sci., 2014(21):22-27.
8	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 Y Q, LI M, ZHANG B, et al.. Progress in application of hyperspectral imaging technology in grain quality detection [J]. J. Chin. Cereals Oils, 2021,36(5)165-173.
9	SUN J F, WANG G X, ZHANG H, et al.. Detection of fat content in peanut kernels based on chemometrics and hyperspectral imaging technology [J/OL]. Infrared Phys. Technol., 2020,105:103226 [2022-10-21]. .
10	TALLADA J G, PALACIOS-ROJAS N, ARMSTRONG P R. Prediction of maize seed attributes using a rapid single kernel near infrared instrument [J]. J. Cereal Sci., 2009,50(3):381-387.
11	芦兵,孙俊,杨宁,等.基于SAGA-SVR 预测模型的水稻种子水分含量高光谱检测[J].南方农业学报,2018,49(11):2342-2348.
	LU B, SUN J, YANG N, et al.. Hyperspectral detection of water content in rice seeds based on Saga-SVR prediction model [J]. J. Southern Agric., 2018,49(11):2342-2348.
12	刘婷,王传堂,唐月异,等.花生自然风干种子维生素E含量近红外分析模型构建[J].山东农业科学,2018,50(6):163-166.
	LIU T, WANG C T, TANG Y Y, et al.. Construction of near-infrared analysis model for vitamin E content in peanut naturally dried seeds [J]. Shandong Agric. Sci., 2018,50(6):163-166.
13	唐月异,王秀贞,刘婷,等.花生自然风干种子蔗糖含量近红外定量分析模型构建[J].山东农业科学,2018,50(6):159-162.
	TANG Y Y, WANG X Z, LIU T, et al.. Establishment of near-infrared quantitative a nalysis model for sucrose content in peanut naturally air-dried seeds [J]. Shandong Agric. Sci.,2018,50(6):159-162.
14	张欣,唐月异,胡东青,等.花生自然风干种子芥酸含量近红外分析模型构建[J].山东农业科学,2018,50(10):138-141.
	ZHANG X, TANG Y Y, HU D Q, et al.. Construction of near infrared analysis model for erucic acid content in peanut naturally dried seeds [J]. Shandong Agric. Sci., 2018,50(10):138-141.
15	吴静珠,申舒,董晶晶,等.AOTF 高光谱成像探测作物氮素含量方法[J].传感器与微系统,2017,36(11):64-66, 69.
	WU J Z, SHEN S, DONG J J, et al.. Detection of nitrogen content in crops by AOTF hyperspectral imaging [J]. Sensors Microsys., 2017,36(11):64-66, 69.
16	蒋蘋,罗亚辉,胡文武,等.基于高光谱的油茶籽内部品质检测最优预测模型研究[J].农机化研究,2015,37(7):56-60.
	JIANG P, LUO Y H, HU W W, et al.. Study on optimal prediction model of camellia oleifera seed internal quality detection based on hyperspectrum [J]. J. Agric. Mechan. Res.,2015,37(7):56-60.
17	郝勇,吴文辉,商庆园,等.山茶油中油酸和亚油酸近红外光谱分析模型[J].光学学报,2019,39(9):381-386.
	HAO Y, WU W H, SHANG Q Y, et al.. Analysis model of oleic acid and linoleic acid in camellia oil by near infrared spectroscopy [J]. Acta Optica Sin.,2019,39(9):381-386.
18	李建国,薛晓梦,张照华,等.单粒花生主要脂肪酸含量近红外预测模型的建立及其应用[J].作物学报,2019,45(12):1891-1898.
	LI J G, XUE X M, ZHANG Z H, et al.. Establishment and application of near-infrared prediction model of main fatty acid content in single peanut [J]. Acta Agron. Sin.,2019,45(12):1891-1898.
19	RAVIKANTH L, CHELLADURAI V, JAYAS D S, et al.. Detection of broken kernels content in Bulk wheat samples using near-infrared imaging [J]. Agric. Res., 2016,5(3):285-292.
20	郭飞,许镇,马宏宏,等.基于PCA的土壤Cd含量高光谱反演模型对比研究[J]. 光谱学与光谱分析, 2021,41(5):1625-1630.
	GUO F, XU Z, MA H H, et al.. Comparison of hyperspectral inversion models for soil Cd content based on PCA [J]. Spectroscopy Spectral Anal., 201,41(5):1625-1630.
21	NAWAR S, MOUAZEN A M. Predictive performance of mobile vis-near infrared spectroscopy for key soil properties at different geographical scale s by using spiking and data mining techniques [J]. Catena, 2017,151:118-129.
22	程介虹,陈争光,张庆华.不同波长选择方法在土壤有机质含量检测中对比研究[J].中国农业科技导报, 2020,22(1):162-170.
	CHENG J H, CHEN Z G, ZHANG Q H. Comparative study of different wavelength selection methods in soil organic matter determination [J]. J. Agric. Sci. Technol., 2020,22(1):162-170.
23	张磊,陈国兴,何学文.基于改进的网格搜索法的SVM参数优化[J].应用科技,2012,39(3):28-31.
	ZHANG L, CHEN G X, HE X W. A parameter optimization method for an SVM based on improved grid search algorithm [J]. Appl. Security, 2012,39(3):28-31.
24	BAIANO A, TERRACONE C, PERI G, et al.. Application of hyperspectral imaging for prediction of physico-chemical and sensory characteristics of table grapes [J]. Comput. Electron. Agric., 2012,87:142-151.
25	PU H, CHEN Z, WANG B, et al.. A novel spatial-spectral similarity measure of dimensionality reduction and classification of hyperspectral imagery [J]. IEEE Trans. Geosci. Remote Sens., 2014,52(11):7008-7022.
26	李维香,张继光,吴万富,等.粮食脂肪酸值测定影响因素分析及优化[J].中国粮油学报, 2021,36(3): 154-160.
	LI W X, ZHANG J G, WU W F, et al.. Analysis and optimization of influencing factors of fatty acid value in grain [J]. J. Chin. Cereals Oils Soc., 201,36(3):154-160.

方法Method	校正集Correction set			预测集Prediction set
方法Method	P_C	R_C	RMSE_C	R_P	RMSE_P	RPD
均值中心化MC	6	0.811	1.281	0.563	1.477	1.532
标准化Autoscales	9	0.644	1.323	0.632	1.543	1.845
标准正态变量交化SNV	9	0.644	1.323	0.632	1.543	1.845
SG平滑Savitzk-golay	5	0.534	2.935	0.601	2.514	1.554
多元散射校正MSC	5	0.765	1.214	0.743	1.421	2.124
移动平均平滑MA	7	0.721	1.645	0.563	1.458	1.542
归一化Normalize	10	0.779	1.911	0.350	2.212	1.143
直接差分二阶求导DDSD	8	0.685	1.412	0.523	1.442	1.723
直接差分一阶求导DDFD	9	0.652	1.321	0.624	2.031	1.586
二阶求导2^nd-derivative	14	0.356	3.451	0.528	1.584	1.295
一阶求导1^st-derivative	12	0.632	1.356	0.634	1.536	1.846

方法Method	校正集Correction set			预测集Prediction set
方法Method	P_C	R_C	RMSE_C	R_P	RMSE_P	RPD
均值中心化MC	6	0.811	1.281	0.563	1.477	1.532
标准化Autoscales	9	0.644	1.323	0.632	1.543	1.845
标准正态变量交化SNV	9	0.644	1.323	0.632	1.543	1.845
SG平滑Savitzk-golay	5	0.534	2.935	0.601	2.514	1.554
多元散射校正MSC	5	0.765	1.214	0.743	1.421	2.124
移动平均平滑MA	7	0.721	1.645	0.563	1.458	1.542
归一化Normalize	10	0.779	1.911	0.350	2.212	1.143
直接差分二阶求导DDSD	8	0.685	1.412	0.523	1.442	1.723
直接差分一阶求导DDFD	9	0.652	1.321	0.624	2.031	1.586
二阶求导2^nd-derivative	14	0.356	3.451	0.528	1.584	1.295
一阶求导1^st-derivative	12	0.632	1.356	0.634	1.536	1.846

主成分 Principal component	贡献率 Contribution rate/%	累计贡献率 Cumulative contribution rate/%
P1	87.77	87.77
P2	9.77	97.54
P3	1.26	98.81
P4	0.27	99.08
P5	0.14	99.23

主成分 Principal component	贡献率 Contribution rate/%	累计贡献率 Cumulative contribution rate/%
P1	87.77	87.77
P2	9.77	97.54
P3	1.26	98.81
P4	0.27	99.08
P5	0.14	99.23

模型 Model	处理方法 Processing method	校正集 Calibration set		预测集 Prediction set
模型 Model	处理方法 Processing method	R_c	RMSE_C	R_p	RMSE_P
SVM	MSC+PCA+SVM	0.844	1.124 4	0.648	3.563 8
	MSC+CARS+SVM	0.796	4.234 7	0.721	3.894 5
	MSC+SPA+SVM	0.772	2.423 5	0.760	2.562 6
ELM	MSC+PCA+ELM	0.774	3.473 4	0.735	2.143 4
	MSC+CARS+ELM	0.894	0.993 4	0.868	1.069 8
	MSC+SPA+ELM	0.763	1.836 2	0.778	1.783 4
LSSVM	MSC+PCA+LSSVM	0.820	1.146 3	0.596	4.233 9
	MSC+CARS+LSSVM	0.742	1.968 2	0.667	3.847 2
	MSC+SPA+LSSVM	0.801	1.804 5	0.791	1.769 8