• EI
  • Scopus
  • 食品科学与工程领域高质量科技期刊分级目录第一方阵T1
  • DOAJ
  • EBSCO
  • 北大核心期刊
  • 中国核心学术期刊RCCSE
  • JST China
  • FSTA
  • 中国精品科技期刊
  • 中国农林核心期刊
  • CA
  • WJCI
  • 中国科技核心期刊CSTPCD
  • 中国生物医学SinoMed
中国精品科技期刊2020
刘诚,赵路路,周松斌,等. 基于高光谱成像技术的陈皮年份快速鉴别[J]. 食品工业科技,2024,45(24):253−262. doi: 10.13386/j.issn1002-0306.2024010229.
引用本文: 刘诚,赵路路,周松斌,等. 基于高光谱成像技术的陈皮年份快速鉴别[J]. 食品工业科技,2024,45(24):253−262. doi: 10.13386/j.issn1002-0306.2024010229.
LIU Cheng, ZHAO Lulu, ZHOU Songbin, et al. Rapid Discrimination of Aging Year of Chenpi Based on Hyperspectral Images[J]. Science and Technology of Food Industry, 2024, 45(24): 253−262. (in Chinese with English abstract). doi: 10.13386/j.issn1002-0306.2024010229.
Citation: LIU Cheng, ZHAO Lulu, ZHOU Songbin, et al. Rapid Discrimination of Aging Year of Chenpi Based on Hyperspectral Images[J]. Science and Technology of Food Industry, 2024, 45(24): 253−262. (in Chinese with English abstract). doi: 10.13386/j.issn1002-0306.2024010229.

基于高光谱成像技术的陈皮年份快速鉴别

Rapid Discrimination of Aging Year of Chenpi Based on Hyperspectral Images

  • 摘要: 陈皮具有较好的经济价值与药用价值,但目前市场上假冒伪劣、以次充好的现象严重。尤其是陈皮陈化年份作为衡量陈皮品质的重要指标,采用人工检测方法准确率与效率较低。为此,本文采用高光谱成像技术结合深度学习方法,建立陈皮陈化年份的快速无损鉴别方法。采集4类不同陈化年份的480个陈皮样本的近红外高光谱数据(波长范围为935.61~1720.23 nm),并采用轻量化卷积网络1D-Rep网络建立分类模型。在此网络基础上,提出基于多层梯度加权类激活映射(M-Grad-CAM)的特征波段选择方法,并建立特征波段分类模型。该方法综合加权多个Rep-block层的梯度生成波段重要性曲线,从而实现融合波段领域相关性与远程相关性的波段重要性指示。为验证方法有效性,采用基于偏最小二乘判别分析(PLS-DA)、随机森林(RF)、支持向量机(SVM)等机器学习方法获得的特征波段作为对比方法。结果表明,1D-Rep全波段光谱模型准确率达到98.55%。在特征波段建模的情况下,采用M-Grad-CAM选取特征波长,基于前9个特征波段建立分类模型准确率可超过90%,在20个特征波段时达到96.82%,准确率显著优于其他对比模型。本研究采用高光谱成像技术,可有效对不同年份的陈皮进行无损鉴别,并为开发便携检测仪器提供方法和理论依据。

     

    Abstract: Chenpi, or sun-dried mandarin orange peel, holds significant economic and medicinal value, yet counterfeit and substandard products are prevalent in the current market. The aging year of Chenpi is a crucial quality indicator, but accurately determining it through manual inspection is challenging. This study proposed a rapid, non-destructive method to discern the aging year of Chenpi by integrating hyperspectral imaging with deep learning. A total of 480 Chenpi samples across four aging years were collected, and their near-infrared hyperspectral data (wavelength range: 935.61~1720.23 nm) were obtained. A lightweight 1D-Rep network was utilized to develop a classification model enhanced by a feature band selection technique using multi-layer gradient-weighted class activation mapping (M-Grad-CAM). This approach evaluates the importance of spectral bands across multiple Rep-block layers, indicating band significance while considering inter-band and remote correlations. To validate the effectiveness of the proposed method, feature bands obtained from machine learning methods such as partial least squares discriminant analysis (PLS-DA), random forest (RF), and support vector machine (SVM) were used as comparative methods. The results showed that, the 1D-Rep full-spectrum model achieved an accuracy of 98.55%. When employing M-Grad-CAM for feature band selection and establishing a classification model based on the first nine feature bands, an accuracy greater than 90% could be achieved in feature band modeling. The accuracy reached 96.82% with 20 feature bands, significantly higher than that of the comparative models. This research effectively distinguishes Chenpi of different years using hyperspectral imaging technology, providing a methodological and theoretical basis for the development of portable detection instruments.

     

/

返回文章
返回