论文部分内容阅读
利用400~1000 nm可见近红外高光谱成像系统对鸡肉嫩度进行快速无损检测研究。采集鸡肉表面的高光谱散射图像,提取样本感兴趣区域反射光谱曲线并用剪切力值表征鸡肉的标准嫩度。以原始光谱和多元散射校正(MSC)预处理光谱数据建立鸡肉嫩度的偏最小二乘回归(PLSR)模型,预处理光谱建立的模型效果更优。基于MSC预处理,采用偏PLS权重系数法结合逐步回归法筛选出了4个特征波长。然后采用PLSR和多元线性回归(MLR)模型分别建立特征波长处光谱反射值和鸡肉嫩度关系的数学模型,优选最佳模型。结果显示:MLR模型预测效果较好,预测相关系数(RP)和均方根误差(RMSEP)分别为0.94和1.97。研究表明:利用可见近红外高光谱成像技术结合多元回归分析法对鸡肉嫩度的快速无损检测是可行的。
Fast and nondestructive testing of chicken tenderness with 400 ~ 1000 nm visible near infrared hyperspectral imaging system. Hyperspectral scatter images of chicken surface were collected, the reflectance spectral curves of the region of interest were extracted and the standard tenderness of chicken meat was characterized by shear force values. The partial least squares regression model (PLSR) of chicken tenderness was established by using the original spectra and the multiple scattering correction (MSC) pretreatment spectral data. The model pretreated by pretreatment was more effective. Based on MSC preprocessing, four characteristic wavelengths were screened by PLS weighted coefficient method combined with stepwise regression. Then the mathematical model of the relationship between the spectral reflectance at the characteristic wavelength and the tenderness of chicken meat was established by using the PLSR and multiple linear regression (MLR) models respectively, and the optimal model was optimized. The results show that MLR model has a good prediction effect, and the prediction correlation coefficient (RP) and root mean square error (RMSEP) are 0.94 and 1.97 respectively. The results show that it is feasible to detect the tenderness of chicken meat by using visible near-infrared hyperspectral imaging combined with multiple regression analysis.