太赫兹光谱转图像及ResNet的陈皮年份鉴别
ChenpiAge Identification Based on Terahertz Spectral Imaging and ResNet
DOI: 10.12677/AIRR.2024.131002, PDF,    科研立项经费支持
作者: 冯明城, 郑成勇*, 查小文, 陈 帅：五邑大学数学与计算科学学院，广东 江门
关键词: 太赫兹光谱；ResNet；陈皮年份鉴别；谱转图像；Terahertz Spectrum； ResNet； Chenpi Age Identification； Spectral Imaging
摘要: 中国陈皮(学名柑橘皮，PCR)，它们具有高药用价值且随时间增长而增值，但是由于各个年份的陈皮的形态类似，难以区分。因此本研究提出使用太赫兹技术结合深度学习方法对陈皮样品进行贮藏年份的鉴别。利用图像特征转换方法，包括Gramian角场(GAF)、Markov过渡场(MTF)和递归图(RP)，将光谱数据成功转化为RGB三维图像数据，构建了ResNet模型，然后将ResNet模型与1D-CNN模型和CNN-LSTM模型进行对比。结果显示，ResNet模型表现最优，准确率达到了0.8035。总之，将太赫兹光谱数据转换为图像数据结合深度学习模型可以有效区分不同贮藏年份的陈皮，为陈皮的年份检测提供了一个有效的方法。
Abstract: Chinese Chenpi, known scientifically as pericpium citri reticulatae (PCR), possess significant medicinal value and appreciate in worth over time. However, the morphological similarities among Chenpi from different years make them difficult to distinguish. Hence, this study proposes the utilization of terahertz technology in conjunction with deep learning methods for identifying the storage years of mandarin peel samples. Through the application of image feature transformation techniques, such as Gramian Angular Field (GAF), Markov Transition Field (MTF), and Recurrence Plot (RP), the spectral data is successfully transformed into RGB three-dimensional image data. Subsequently, a ResNet model is constructed and compared to 1D-CNN and CNN-LSTM models. The results indicate that the ResNet model outperforms, achieving an accuracy rate of 0.8035. In conclusion, converting terahertz spectral data into image data, in combination with deep learning models, offers an effective means of distinguishing Chenpi of different storage years, thereby providing an efficient approach to mandarin Chenpi year detection.
文章引用：冯明城, 郑成勇, 查小文, 陈帅. 太赫兹光谱转图像及ResNet的陈皮年份鉴别[J]. 人工智能与机器人研究, 2024, 13(1): 9-18. https://doi.org/10.12677/AIRR.2024.131002

参考文献

[1] Zhou, W., Chen, Z., Lu, A., et al. (2021) Systems Pharmacology-Based Strategy to Explore the Pharmacological Mechanisms of Citrus Peel (Chenpi) for Treating Complicated Diseases. The American Journal of Chinese Medicine, 49, 391-411. [Google Scholar] [CrossRef] 
[2] Wang, F., Chen, L., Chen, S., et al. (2021) Microbial Biotransformation of Pericarpium Citri Reticulatae (PCR) by Aspergillus niger and Effects on Antioxidant Activity. Food Science & Nutrition, 9, 855-865. [Google Scholar] [CrossRef] [PubMed]
[3] Li, P., Zhang, X., Zheng, Y., et al. (2021) A Novel Method for the Nondestructive Classification of Different—Age Citri Reticulatae Pericarpium Based on Data Combination Technique. Food Science & Nutrition, 9, 943-951. [Google Scholar] [CrossRef] [PubMed]
[4] Wang, J., Liu, Y., Chen, H., et al. (2013) Chemical Changeability of Essential Oils in Chenpi and Qingpi from the Same Origin by Gas Chromatography-Mass Spectrometry Compiled with Automated Mass Spectral Deconvolution and Identification System. Asian Journal of Chemistry, 25, 6434-6442.
[5] Huang, L., Sun, D.W., Pu, H., et al. (2023) Nanocellulose-Based Polymeric Nanozyme as Bioinspired Spray Coating for Fruit Preservation. Food Hydrocolloids, 135, Article ID: 108138. [Google Scholar] [CrossRef] 
[6] Luo, M., Luo, H., Hu, P., et al. (2018) Evaluation of Chemical Components in Citri Reticulatae Pericarpium of Different Cultivars Collected from Different Regions by GC-MS and HPLC. Food Science & Nutrition, 6, 400-416. [Google Scholar] [CrossRef] [PubMed]
[7] Li, D., Zhu, Z. and Sun, D.W. (2022) Visualization and Quantification of Content and Hydrogen Bonding State of Water in Apple and Potato Cells by Confocal Raman Microscopy: A Comparison Study. Food Chemistry, 385, Article ID: 132679. [Google Scholar] [CrossRef] [PubMed]
[8] Duan, L., Guo, L., Dou, L.L., et al. (2016) Discrimination of Citrus reticulata Blanco and Citrus reticulata ‘Chachi’ by Gas Chromatograph-Mass Spectrometry Based Metabolomics Approach. Food Chemistry, 212, 123-127. [Google Scholar] [CrossRef] [PubMed]
[9] Cho, H.E., Ahn, S.Y., Kim, S.C., et al. (2014) Determination of Flavonoid Glycosides, Polymethoxyflavones, and Coumarins in Herbal Drugs of Citrus and Poncirus Fruits by High Performance Liquid Chromatography—Electrospray ionization/Tandem Mass Spectrometry. Analytical Letters, 47, 1299-1323. [Google Scholar] [CrossRef] 
[10] Sun, Y.J. and Wang, J.F. (2022) Research Progress of Terahertz Time-Domain Spectroscopy in Food, Drugs, and Environment. Laser & Optoelectronics Progress, 59, Article ID: 1600003. [Google Scholar] [CrossRef] 
[11] 孙一健, 王继芬. 太赫兹时域光谱技术在食品、药品和环境领域中的应用研究进展[J]. 激光与光电子学进展, 2022, 59(16): 12-21.
[12] El Haddad, J., Bousquet, B., Canioni, L., et al. (2013) Review in Terahertz Spectral Analysis. TrAC Trends in Analytical Chemistry, 44, 98-105. [Google Scholar] [CrossRef] 
[13] Pu, H., Yu, J., Sun, D.W., et al. (2023) Distinguishing Pericarpium Citri Reticulatae of Different Origins Using Terahertz Time-Domain Spectroscopy Combined with Convolutional Neural Networks. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 299, Article ID: 122771. [Google Scholar] [CrossRef] [PubMed]
[14] Liu, Y., Pu, H., Li, Q., et al. (2023) Discrimination of Pericarpium Citri Reticulatae in Different Years Using Terahertz Time-Domain Spectroscopy Combined with Convolutional Neural Network. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 286, Article ID: 122035. [Google Scholar] [CrossRef] [PubMed]
[15] Liu, W., Zhang, R., Ling, Y., et al. (2020) Automatic Recognition of Breast Invasive Ductal Carcinoma Based on Terahertz Spectroscopy with Wavelet Packet Transform and Machine Learning. Biomedical Optics Express, 11, 971-981. [Google Scholar] [CrossRef] 
[16] Chen, Z., Zhang, Z., Zhu, R., et al. (2015) Application of Terahertz Time-Domain Spectroscopy Combined with Chemometrics to Quantitative Analysis of Imidacloprid in Rice Samples. Journal of Quantitative Spectroscopy and Radiative Transfer, 167, 1-9. [Google Scholar] [CrossRef] 
[17] Yang, R., Li, Y., Qin, B., et al. (2022) Pesticide Detection Combining the Wasserstein Generative Adversarial Network and the Residual Neural Network Based on Terahertz Spectroscopy. RSC Advances, 12, 1769-1776. [Google Scholar] [CrossRef] 
[18] Tang, M., Xia, L., Wei, D., et al. (2020) Rapid and Label-Free Metamaterial-Based Biosensor for Fatty Acid Detection with Terahertz Time-Domain Spectroscopy. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 228, Article ID: 117736. [Google Scholar] [CrossRef] [PubMed]
[19] Bland, M. (2015) Estimating Mean and Standard Deviation from the Sample Size, Three Quartiles, Minimum, and Maximum. International Journal of Statistics in Medical Research, 4, 57. [Google Scholar] [CrossRef] 
[20] Wang, Z. and Oates, T. (2015) Encoding Time Series as Images for Visual Inspection and Classification Using Tiled Convolutional Neural Networks. Workshops at the Twenty-Ninth AAAI Conference on Artificial Intelligence, Austin, 25-30 January 2015, 40-46.
[21] Eckmann, J.P., Kamphorst, S.O. and Ruelle, D. (1995) Recurrence Plots of Dynamical Systems. World Scientific Series on Nonlinear Science Series A, 16, 441-446. [Google Scholar] [CrossRef] 
[22] Chawla, N.V., Bowyer, K.W., Hall, L.O., et al. (2002) SMOTE: Synthetic Minority Over-Sampling Technique. Journal of Artificial Intelligence Research, 16, 321-357. [Google Scholar] [CrossRef] 
[23] Khalifa, N.E., Loey, M. and Mirjalili, S. (2022) A Comprehensive Survey of Recent Trends in Deep Learning for Digital Images Augmentation. Artificial Intelligence Review, 55, 2351-2377.
[24] He, K., Zhang, X., Ren, S., et al. (2016) Deep Residual Learning for Image Recognition. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, 27-30 June 2016, 770-778. [Google Scholar] [CrossRef] 
[25] Aksan, F., Li, Y., Suresh, V., et al. (2023) CNN-LSTM vs. LSTM-CNN to Predict Power Flow Direction: A Case Study of the High-Voltage Subnet of Northeast Germany. Sensors, 23, 901. [Google Scholar] [CrossRef] [PubMed]
[26] Li, L., Doroslovački, M. and Loew, M.H. (2020) Approximating the Gradient of Cross-Entropy Loss Function. IEEE Access, 8, 111626-111635. [Google Scholar] [CrossRef] 

为你推荐