农业大数据学报 ›› 2019, Vol. 1 ›› Issue (2): 51-63.doi: 10.19788/j.issn.2096-6369.190205

• 专题-植物表型组学 • 上一篇    下一篇

基于高光谱的水稻精米品质参数测量技术研究

戴国新1,2,陈国兴3,杨万能1,2,3,冯慧1,2,*()   

  1. 1.华中农业大学国家农作物分子技术育种中心,作物遗传改良国家重点实验室,武汉 430070
    2.农业生物信息湖北省重点实验室,华中农业大学信息学院,武汉 430070
    3.华中农业大学植物科学技术学院,武汉 430070
  • 收稿日期:2019-04-12 出版日期:2019-06-26 发布日期:2019-08-21
  • 通讯作者: 冯慧 E-mail:fenghui@mail.hzau.edu.cn
  • 作者简介:戴国新,男,硕士,研究方向:作物高光谱;Email: 1904536143@qq.com
  • 基金资助:
    自然科学基金青年科学基金(21800305);华中农业大学自主科技创新基金(2662017QD044);国家重点研发计划“七大农作物育种"重点专项(2018YFD0100900)

Measurement Technology of Quality Parameters of Rice Grain Based on Hyperspectral Imaging on the Visible-near Infrared

Guoxin Dai1,2,Guoxin Chen3,Wanneng Yang1,2,3,Hui Feng1,2,*()   

  1. 1.National Key Laboratory of Crop Genetic Improvement, National Center of Plant Gene Research, Huazhong Agricultural University, Wuhan 430070
    2.Agricultural Bioinformatics Key Laboratory of Hubei Province, and College of Informations, Huazhong Agricultural University, Wuhan 430070
    3.College of Plant Science & Technology, Huazhong Agricultural University, Wuhan 430070
  • Received:2019-04-12 Online:2019-06-26 Published:2019-08-21
  • Contact: Hui Feng E-mail:fenghui@mail.hzau.edu.cn

摘要:

【目的】水稻精米直链淀粉和蛋白质含量,是衡量稻米加工特性的重要参数。本研究旨在使用高光谱技术来无损检测其含量。【方法】以水稻核心种质资源材料中106个材料为研究对象,使用可见光-近红外高光谱成像技术获取各波段下光谱指数。【结果】高光谱数据经过数据重整、图像提取及分析等一系列操作后,可以得到初级指数和复杂指数。初级指数与直链淀粉和蛋白质含量模型的R2分别为0.823和0.837,五倍交叉验证结果显示模型稳定性良好。同时,通过复杂指数与直链淀粉和蛋白质含量的相关系数结果也可以看出,500-800nm区域间,高光谱复杂指数与直链淀粉和蛋白质含量相关系数较高。【结论】基于高光谱成像技术可以同时准确估测精米直链淀粉和蛋白质含量,为后续水稻种质资源批量品质分析打下研究基础。

关键词: 高光谱, 水稻, 精米, 直链淀粉, 蛋白质, 线性回归, 相关系数, 植物表型组学

Abstract:

[Objective] The objective of this study is to develop a novel non-destructive and rapid optical method for the measurement of amylose and protein content in rice, which can be used to measure the processing characteristics of rice after cooking. [Method] A new measuring technique based on the acquisition and analysis of near-infrared hyperspectral images was developed to acquire hyperspectral indices within various wavelengths of 106 polished rice accessions in rice core germplasm resources. [Result]The model for stepwise linear regression analysis showed that the R2 of the amylose and protein content models were 0.823 and 0.837, respectively, and the five-fold cross-validation results demonstrated the stability of the models. In the wavelength range of 500-800nm, the correlation coefficient between them was higher, meanwhile, the more the index varied, the less information was available, which showed that among massive spectral data, the information concentration area was very limited.[Conclusion] The novel method based on hyperspectral imaging technology can estimate amylose and protein content non-destructively and accurately, which is the basis for further analysis of the grain quality of generous core rice accessions.

Key words: hyperspectral, rice, polished rice, amylose content, protein content, linear regression, correlation coefficient, plant phenom

中图分类号: 

  • S-3