• Spectroscopy and Spectral Analysis
  • Vol. 41, Issue 8, 2572 (2021)
yao FANG1、*, Tian-hua XIE2、2;, Wei GUO1、1;, Xue-bing BAI1、1;, and Xin-xing LI1、1; *;
Author Affiliations
  • 11. Beijing Laboratory of Food Quality and Safety, College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China
  • 22. College of Engineering, China Agricultural University, Beijing 100083, China
  • show less
    DOI: 10.3964/j.issn.1000-0593(2021)08-2572-06 Cite this Article
    yao FANG, Tian-hua XIE, Wei GUO, Xue-bing BAI, Xin-xing LI. On-Line Fast Detection Technology of Chilled Fresh Meat Quality Based on Hyperspectral and Multi-Parameter[J]. Spectroscopy and Spectral Analysis, 2021, 41(8): 2572 Copy Citation Text show less
    Curve of different texture parameters changing with time(a): Hardness over time; (b): Resilience of hardness over time; (c): Chewiness of hardness over time;(d): Cohesivencess of hardness time; (e): Stalemate of hardness over time; (f): Resilience of hardness over time
    Fig. 1. Curve of different texture parameters changing with time
    (a): Hardness over time; (b): Resilience of hardness over time; (c): Chewiness of hardness over time;(d): Cohesivencess of hardness time; (e): Stalemate of hardness over time; (f): Resilience of hardness over time
    Raw spectral curve of chilled beef
    Fig. 2. Raw spectral curve of chilled beef
    Optimal wavelength selected by SPA
    Fig. 3. Optimal wavelength selected by SPA
    Regression model of TPA(a): Scatter plots of true and predicted values of cohesivencess;(b): Scatter plots of ture and predicted values of resilience
    Fig. 4. Regression model of TPA
    (a): Scatter plots of true and predicted values of cohesivencess;(b): Scatter plots of ture and predicted values of resilience
    质构特性训练集预测集
    R2RSDR2RSD
    粘聚性0.597 60.095 20.554 70.087 3
    回复性0.562 30.117 60.507 40.214 8
    胶着度0.303 60.303 00.242 10.254 2
    咀嚼度0.272 00.310 40.255 50.320 6
    弹性0.440 60.043 00.201 00.050 3
    硬度0.464 60.187 20.309 50.260 5
    Table 1. Results of texture characteristic parameters
    预处理
    方法
    主成
    分数
    训练集结果预测集结果
    R2RSDR2RSD
    30.5970.095 20.554 70.087 3
    KS30.592 80.097 10.596 60.081 8
    SPXY30.595 40.094 50.769 30.093 4
    Table 2. Results of different sample partitioning methods
    质构参数预处理方法预测集结果
    R2RSD
    粘聚性D1st0.720 60.097 3
    D2st0.750 60.089 0
    SNV0.814 30.122 9
    MSC0.739 70.112 9
    回复性D1st0.724 40.171 0
    D2st0.867 40.099 4
    SNV0.819 90.175 1
    MSC0.731 50.123 4
    Table 3. Results of different pretreatment method
    质构参数波长/nm预测集结果
    R2RSD
    粘聚度983 1 269 1 429 1 6600.792 40.109 2
    回复性1 061 1 355 1 452 1 6970.739 90.175 9
    Table 4. Modeling results after feature wavelength extraction
    TPA预测集结果
    R2RMSEP
    粘聚性0.879 80.396 6
    回复性0.880 60.220 3
    Table 5. Modeling results based on PLSR
    TPA预测集结果
    R2RMSEP
    粘聚性0.728 80.383 2
    回复性0.766 00.284 3
    Table 6. Modeling results based on PCR
    yao FANG, Tian-hua XIE, Wei GUO, Xue-bing BAI, Xin-xing LI. On-Line Fast Detection Technology of Chilled Fresh Meat Quality Based on Hyperspectral and Multi-Parameter[J]. Spectroscopy and Spectral Analysis, 2021, 41(8): 2572
    Download Citation