Hyperspectral Imaging-Based Simultaneous Assessment of Processing Quality of Chilled Pork
HE Jiaxin, GUO Chuanyu, LIU Yuling, WANG Yawen, HAN Ge, PAN Jinfeng, LI Shengjie
1. School of Food Science and Technology, Dalian Polytechnic University, Dalian 116034, China; 2. National Engineering Research Center of Seafood, Dalian 116034, China; 3. State Key Laboratory of Seafood Processing and Safety Control, Dalian 116034, China
摘要本研究旨在建立一种基于高光谱成像技术的快速检测方法，用于同步检测冷鲜猪肉的质构特性和保水性。采集冷鲜猪肉的可见-近红外（400～1 050 nm）和近红外（900～1 700 nm）2 个波段的高光谱图像，分别对比3 种光谱预处理方法（Savitzky-Golay（S-G）平滑、一阶导数、二阶导数）以及2 种特征波长选择方法（连续投影法、二维相关光谱）对偏最小二乘回归（partial least squares regression，PLSR）模型预测性能的影响。结果表明，采用近红外波段数据建立的PLSR模型，其预测决定系数（coefficient of determination for prediction，R2p）和校正决定系数（coefficient of determination for calibration，R2c）更接近1，校正均方根误差和预测均方根误差也更小，整体性能优于基于可见-近红外波段数据建立的PLSR模型。在光谱预处理方法中，基于二阶导数预处理的近红外-PLSR模型表现最佳，其R2c介于0.848～0.932之间，R2p介于0.648～0.829之间。此外，无论采用连续投影法还是二维相关光谱法进行特征波长提取，均会导致PLSR模型预测能力的显著下降。综上所述，直接采用全谱数据并配合二阶导数预处理所建立近红外高光谱模型，为实现冷鲜猪肉质构与保水性的快速、无损检测提供了可靠的技术方案。
Abstract：This study aimed to establish a rapid method based on hyperspectral imaging for the simultaneous assessment of texture characteristics and water-holding capacity in chilled pork. The visible-near-infrared (VIS-NIR, 400–1 050 nm) and near-infrared (NIR, 900–1 700 nm) spectra of pork samples were acquired. Three spectral preprocessing methods—Savitzky-Golay (S-G) smoothing, first derivative (1st-der), and second derivative (2nd-der)—as well as two feature wavelength selection methods—successive projections algorithm (SPA) and two-dimensional correlation spectroscopy (2D-COS)—were compared for their effects on the predictive performance of the partial least squares regression (PLSR) model. The results showed that the model developed using the NIR spectral range exhibited prediction coefficient of determination (R2p) and calibration coefficient of determination (R2 c) values closer to 1, along with lower root mean square error of calibration and root mean square error of prediction, indicating better overall performance compared to the model based on the VISNIR data. Among the spectral preprocessing methods, the best one was 2nd-der, and the 2nd-der-NIR-PLSR model exhibited the best performance with R2c ranging from 0.848 to 0.932 and R2p ranging from 0.648 to 0.829. Additionally, it was found that both SPA and 2D-COS for feature wavelength extraction led to a significant decline in the predictive ability of the PLSR model. In conclusion, the model established based on the full spectral data with 2nd-der preprocessing provides a reliable technical solution for the rapid and non-destructive detection of texture and water-holding capacity in chilled pork.
何佳欣，郭传宇，刘玉玲，王雅雯，韩 格，潘锦锋，李胜杰. 基于高光谱技术同步检测冷鲜猪肉的加工品质[J]. 肉类研究, 2026, 40(10): 88-93 https://doi.org/10.7506/rlyj1001-8123-20251107-337
HE Jiaxin, GUO Chuanyu, LIU Yuling, WANG Yawen, HAN Ge, PAN Jinfeng, LI Shengjie. Hyperspectral Imaging-Based Simultaneous Assessment of Processing Quality of Chilled Pork. Meat Research, 2026, 40(10): 88-93 https://doi.org/10.7506/rlyj1001-8123-20251107-337
2026, Vol. 40 
