高光谱成像技术结合深度学习的藕粉识别和掺假检测

doi:10.3964/j.issn.1000-0593(2025)06-1759-09

摘要
参考文献
相关文章 (15)

全文: PDF (13001 KB)
输出: BibTeX | EndNote (RIS)

摘要：藕粉营养价值高，工艺复杂，一些不法商家受到利益的驱使，利用廉价的普通淀粉冒充藕粉或在藕粉中掺入普通淀粉。传统的藕粉真伪检查方法耗时耗力，具有破坏性。高光谱成像技术凭借其快速、无损且精确的优点在食品安全检测领域得到广泛应用。因此，为了准确区分藕粉和其他普通淀粉并识别掺假藕粉，提出了一种高光谱成像技术结合深度学习的快速鉴别藕粉真伪的方法。利用高光谱成像技术采集900~1 700 nm波段范围内的纯藕粉、四种普通淀粉以及掺假淀粉的高光谱图像。在纯藕粉和四种普通淀粉的高光谱图像中划分若干个感兴趣区域（ROI），计算每个ROI的平均反射率作为构建分类模型的原始光谱数据。去除掉原始光谱前后受噪音影响的异常波段，保留了940~1 675 nm之间的443个波段。接着通过孤立森林（IF）算法剔除掉光谱数据中的异常数据。为提高模型训练效率，采用竞争性自适应重加权算法（CARS）、自助软收缩算法（BOSS）和通道注意力模块（CAMM）三种方法分别从443个波段中提取出45、32和12个特征波长。基于提取出的特征波长的光谱数据，构建了偏最小二乘判别（PLS-DA）分类模型，其中CAMM-PLS-DA模型识别效果最好，测试集准确率达到了95.25%。为了确定最佳分类模型，基于CAMM提取不同特征波长数目下的光谱数据，建立PLS-DA、支持向量机（SVM）和卷积神经网络（CNN）分类模型，其中CAMM-CNN模型的分类性能最好，测试集准确率最高达到了99.69%。为进一步检验CAMM-CNN模型对掺假藕粉的鉴别能力，将掺假藕粉高光谱图像所有像素点的光谱数据输入到训练好的CAMM-CNN模型中进行判别，从可视化图像看出，模型成功识别出掺假藕粉中的多种普通淀粉。研究结果表明，高光谱成像技术结合深度学习方法可以有效地应用于藕粉的真伪鉴别，这为打击藕粉掺假行为和保障藕粉安全提供了一种新的检测手段。

关键词：藕粉；掺假；高光谱成像技术；深度学习；通道注意力机制模块；卷积神经网络

Abstract：Lotus root starch is highly nutritious, and its production process is complex. Some unscrupulous businessmen, driven by profit, adulterate lotus root starch with cheaper common starch or mix common starch into lotus root starch. Traditional methods for authenticating lotus root starch are time-consuming, labor-intensive, and destructive. Hyperspectral imaging technology, with its advantages of rapid, non-destructive, and accurate, has been widely applied in food safety detection. Therefore, this study proposes a method for quickly distinguishing between lotus root starch and other common starches, as well as identifying adulterated lotus root starch, by combining hyperspectral imaging technology with deep learning. Hyperspectral images of pure lotus root starch, four types of common starches, and adulterated starches were collected in the wavelength range of 900~1 700 nm using hyperspectral imaging technology. Several regions of interest (ROI) were delineated in the hyperspectral images of pure lotus root starch and the four common starches, and the average reflectance of each ROI was calculated from the original spectral data to build classification models. Abnormal bands affected by noise at the beginning and end of the original spectra were removed, leaving 443 bands between 940 and 1 675 nm. Outliers in the spectral data were then eliminated using the Isolation Forest (IF) algorithm. To enhance model training efficiency, the Competitive Adaptive Reweighted Sampling (CARS), Bootstrapping Soft Shrinkage (BOSS), and Channel Attention Mechanism Module (CAMM) were employed to extract 45, 32, and 12 feature wavelengths from the 443 bands, respectively. Partial Least Squares Discriminant Analysis (PLS-DA) classification models were constructed based on the spectral data after feature wavelength extraction, with the CAMM-PLS-DA model showing the best recognition effect, achieving an accuracy of 95.25% in the test set. To determine the optimal classification model, PLS-DA, Support Vector Machine (SVM), and Convolutional Neural Network (CNN) classification models were established using spectral data with different numbers of feature wavelengths extracted by CAMM. The CAMM-CNN model exhibited the best classification performance, with a highest accuracy of 99.69% in the test set. To further verify the ability of the CAMM-CNN model to distinguish adulterated lotus root starch, the spectral data of all pixel points in the hyperspectral images of adulterated lotus root starch were input into the trained CAMM-CNN model for discrimination. Visualization images showed that the model successfully identified various types of common starches in the adulterated lotus root starch. The results indicate that the combination of hyperspectral imaging technology and deep learning methods can effectively be applied to the authentication of lotus root starch, providing a new detection approach to combat the adulteration of lotus root starch and ensure its safety.

Key words：Lotus root starch; Adulteration; Hyperspectral imaging technology; Deep learning; Channel attention mechanism module; Convolutional neural network

收稿日期: 2024-05-07 修订日期: 2025-01-13

中图分类号:

TS237

基金资助: 国家自然科学基金项目（32372310），四川省科技厅项目（2023YFS0451），酿酒生物技术及应用四川省重点实验开放课题项目(NJ2022-08)和四川轻化工大学研究生创新基金项目(Y2023098)资助

通讯作者: 胡新军 E-mail: suse2021@126.com

作者简介: 彭健恒，2000年生，四川轻化工大学机械工程学院硕士研究生 e-mail: 1198751028@qq.com

引用本文:

彭健恒，胡新军，张嘉洪，田建平，陈满骄，黄丹，罗惠波. 高光谱成像技术结合深度学习的藕粉识别和掺假检测[J]. 光谱学与光谱分析, 2025, 45(06): 1759-1767.
PENG Jian-heng, HU Xin-jun, ZHANG Jia-hong, TIAN Jian-ping, CHEN Man-jiao, HUANG Dan, LUO Hui-bo. Identification and Adulteration Detection of Lotus Root Starch Using Hyperspectral Imaging Technology Combined With Deep Learning. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2025, 45(06): 1759-1767.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2025)06-1759-09 或 https://www.gpxygpfx.com/CN/Y2025/V45/I06/1759

[1] Zhu F. Food Hydrocolloids, 2017, 63: 332.
[2] Niu X, Zhao Z, Jia K, et al. Food Chemistry, 2012, 133(2): 592.
[3] Chen G, Zhu M, Guo M. Critical Reviews in Food Science and Nutrition, 2019, 59(sup1): S189.
[4] Guo Y, Sun K, Cheng Y, et al. European Food Research and Technology, 2022, 248(4): 1117.
[5] Zhang L, Wang P, Li S, et al. Molecules, 2023, 28(5): 2016.
[6] Chepngeno J, Imathiu S, Owino W O, et al. Food Chemistry, 2022, 390: 133108.
[7] Liu Z, Li X, Zhao Y, et al. Microscopy Research and Technique, 2022, 85(7): 2428.
[8] Zhang X, Qiao C, Fu S, et al. Journal of Dairy Science, 2022, 105(6): 4749.
[9] Wang J, Liu Y, Li J, et al. International Journal of Food Properties, 2022, 25(1): 1203.
[10] HAN Jian-xun, CHEN Ying, WU Ya-jun, et al(韩建勋, 陈颖, 吴亚君, 等). Journal of Chinese Institute of Food Science and Technology(中国食品学报), 2019, 19(2): 291.
[11] Bai X, Yuan C, Dong J, et al. Journal of Food Composition and Analysis, 2023, 123: 105576.
[12] Ndlovu P F, Magwaza L S, Tesfay S Z, et al. Food Research International, 2022, 157: 111198.
[13] Liu J, Wen Y, Dong N, et al. Food Chemistry, 2013, 141(3): 3103.
[14] WANG Yan, FU Qi, LI Ying, et al(王燕, 付琪, 李颖, 等). Journal of Food Safety & Quality(食品安全质量检测学报), 2022, 13(15): 5026.
[15] Fu X, Chen J, Fu F, et al. Biosystems Engineering, 2020, 190: 120.
[16] Chen X, Jiao Y, Liu B, et al. Food Chemistry, 2022, 386: 132774.
[17] Jiang X, Bu Y, Han L, et al. Food Control, 2023, 150: 109740.
[18] Hashemi-Nasab F S, Talebian S, Parastar H. Microchemical Journal, 2023, 185: 108203.
[19] Khan M H, Saleem Z, Ahmad M, et al. Neural Computing and Applications, 2021, 33(21): 14507.
[20] Faqeerzada M A, Lohumi S, Kim G, et al. Sensors, 2020, 20(20): 5855.
[21] Shao Y, Liu Y, Xuan G, et al. Infrared Physics & Technology, 2022, 127: 104403.
[22] ZHANG Jia-hong, HE Lin, HU Xin-jun, et al(张嘉洪, 何林, 胡新军, 等). Journal of Food Safety & Quality(食品安全质量检测学报), 2023, 14(20): 209.
[23] Hu Y, Kang Z. Molecules, 2022, 27(4): 1196.
[24] Li H, Liang Y, Xu Q, et al. Analytica Chimica Acta, 2009, 648(1): 77.
[25] Deng B C, Yun Y H, Cao D S, et al. Analytica Chimica Acta, 2016, 908: 63.
[26] Cai Y, Liu X, Cai Z. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(3): 1969.
[27] FANG Jian, YANG Jin-xiang, XIAO Liang(方健, 杨劲翔, 肖亮). Acta Electronica Sinica(电子学报), 2024, 52(1): 201.
[28] YAN Pu-ti, QIU Shi, YUE Cheng-fei(阎菩提, 邱实, 岳程斐). Journal of Nanjing University of Aeronautics and Astronautics(南京航空航天大学学报), 2023, 55(6): 956.
[29] KOU Ze-kun, CHEN Guo-tong, LI Si-yu, et al(寇泽坤, 陈国通, 李思雨, 等). Food Science(食品科学), 2024, 45(1): 254.
[30] MA Ling-kai, ZHU Shi-ping, MIAO Yu-jie, et al(马羚凯, 祝诗平, 苗宇杰, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2022, 42(4): 1222.
[31] Qin Y, Qiu J, Tang N, et al. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2024, 309: 123854.
[32] Han L, Tian J, Huang Y, et al. Journal of Food Composition and Analysis, 2024, 125: 105785.
[33] He H J, Wang Y, Wang Y, et al. International Journal of Biological Macromolecules, 2023, 242(Pt1): 124748.
[34] Liu L, Zareef M, Wang Z, et al. Food Chemistry, 2023, 412: 135505.
[35] Yun Y H, Li H D, Deng B C, et al. TrAC Trends in Analytical Chemistry, 2019, 113: 102.