基于偏振高光谱成像的牛肉品质多参数预测

doi:10.3964/j.issn.1000-0593(2025)07-1820-07

Abstract
References
Related Citation (15)

Download: PDF (8817 KB)
Export: BibTeX | EndNote (RIS)

Abstract Because the traditional method of beef quality detection has the disadvantage of destroying samples, it is impossible to carry out multiple tests on the same sample, and the operation is complicated and the test results are lagging, which is difficult to meet the needs of modern food safety detection for non-destructive and rapid response. To solve the problems of complex operation, great destructiveness, and lag of traditional detection methods, polarization hyperspectral imaging detection technology has been widely used in beef quality detection with its advantages of high precision, fast response, and non-destructiveness. This technology can capture the polarization and spectral information of light, not only has the advantages of hyperspectral imaging technology, non-destructive in-depth analysis of the internal quality of beef, at the same time, the technology introduces polarization imaging technology, can inhibit the influence of the surrounding light environment factors on the spectrum of beef, quickly obtain more accurate results, effectively avoiding the shortcomings of traditional methods. This paper used the wavelength range of 900~1 700 nm to compose a near-infrared polarization highlights like technology, combined with high polarization spectral data and a convolutional neural network, to build a more robust quality parameter prediction model for the first time. First, the hyperspectral data of beef samples were collected without polarization and at 0°, 45°, 90°, and 135° polarization angles, respectively, and the regions of interest were extracted. The samples' color parameters (L^*, a^*, b^*) and texture parameters (hardness, adhesiveness, and cohesiveness) were collected. Secondly, the successive projections algorithmwas used to extract the samples' corresponding spectral characteristic wavelengths with unpolarization and different polarization models. Finally, multiple linear regression and convolutional neural network methods were used to construct the multi-parameter prediction model for beef quality. The results show that the prediction accuracy of multiple parameters in polarization mode is better than that without polarization, and the overall prediction effect of the CNN model at 90° polarization Angle is the best. The determination coefficients of parameters L^*, a^*, b^*, hardness, adhesiveness, and cohesiveness were 0.882, 0.905, 0.949, 0.692, 0.671 and 0.911, and the root-mean-square errors of prediction were 0.820, 0.562, 0.461, 3 889.713, 89.746, and 0.027. Compared with unpolarization, the prediction accuracy of the above 6 parameters is at least 13.1% higher, which verifies the feasibility of the convolutional neural network prediction model combined with near-infrared polarization hyperspectral imaging technology in meat nondestructive testing, and provides a new technical idea and method for further meeting the needs of modern food safety testing.

Key words：Hyperspectral; Polarization; Beef; Prediction model; Color; Texture

Received: 2024-11-26 Accepted: 2025-03-02

ZTFLH:

O433

Corresponding Authors: LI Ya-hong^1*, LI De-yang², ZHANG Yun-cui¹, ZHAO Yu³ E-mail: liyahong@dlpu.edu.cn

Cite this article:

SONG Ya-fang,BU Xiang-tao,LI Na, et al. Multi-Parameter Prediction of Beef Quality Based on Polarized Hyperspectral Imaging[J]. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2025, 45(07): 1820-1826.

URL:

https://www.gpxygpfx.com/EN/10.3964/j.issn.1000-0593(2025)07-1820-07 OR https://www.gpxygpfx.com/EN/Y2025/V45/I07/1820

[1] XU Hui-lin, WEI Jia, ZHANG Xiang, et al(徐慧琳, 魏嘉, 张翔, 等). The Chinese Livestock and Poultry Breeding(中国畜禽种业), 2023, 19(5): 140.
[2] TONG Xiao-bo(佟晓波). Brand and Standardization(品牌与标准化), 2024, (4): 157.
[3] MA Xue-long, XIE Jin-xin, JIANG Nan(马学龙, 解金鑫, 姜楠). Quality and Certification(质量与认证), 2024, (6): 45.
[4] YANG Xiao-ting, LIU Hao, LIU Nan, et al(阳晓婷, 刘浩, 刘楠, 等). Science and Technology of Food Industry(食品工业科技), 2024, 45(11): 37.
[5] Menesatti P, Zanella A, D'Andrea S, et al. Food and Bioprocess Technology, 2009, 2(3): 308.
[6] Wu J H, Peng Y K, Li Y Y, et al. Journal of Food Engineering, 2012, 109: 267.
[7] ElMasry G, Sun D W, Allen P. Journal of Food Engineering, 2012, 110 (1): 127.
[8] YANG Dong, LU An-xiang, WANG Ji-hua(杨东, 陆安祥, 王纪华). Modern Food Science and Technology(现代食品科技), 2017, 33(9): 257.
[9] YU Wen-jie, WANG Cai-xia, QIAO Lu, et al(禹文杰, 王彩霞, 乔芦, 等). Journal of Optoelectronics·Laser(光电子·激光), 2020, 31(3): 326.
[10] WANG Cai-xia, HE Zhi-wu, WU Long-guo, et al(王彩霞, 何智武, 吴龙国, 等). Chinese Journal of Luminescence(发光学报), 2019, 40(4): 520.
[11] FANG Yao, XIE Tian-hua, GUO Wei, et al(方瑶, 谢天铧, 郭渭, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2021, 41(8): 2572.
[12] YAN Chang-xiang, ZHANG Yuan, BO Jian, et al(颜昌翔, 张源, 泊建, 等). Optics and Precision Engineering(光学精密工程), 2024, 32(14): 2141.
[13] YU Yang, ZHANG Jue, TIAN Hai-qing, et al(于洋, 张珏, 田海清, 等). Journal of Agricultural Science and Technology(中国农业科技导报), 2021, 23(12): 101.
[14] LI Zhi-gang, JIA Ce, WANG Xiao-wen, et al(李志刚, 贾策, 王晓闻, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2016, 32(16): 286.
[15] Park S, Yang M, Yim D G, et al. Journal of Food Engineering, 2023, 350: 111500.
[16] Ren Y Q, Fu Y, Sun D W. Food Chemistry, 2023, 428: 136753.
[17] Cen H, He Y. Trends in Food Science & Technology, 2007, 18(2): 72.
[18] Prieto N, Roehe R, Lavín P, et al. Meat Science, 2009, 83(2): 175.