基于多光谱图像技术的冷鲜羊肉新鲜度检测

Freshness Detection of Cold Fresh Lamb Based on Multispectral Image Technology

利用多光谱(320~900 nm)成像技术对冷鲜羊肉建立基于多光谱数据的预测模型,实现对冷鲜羊肉的快速无损检测。采用光谱仪扫描冷鲜羊肉,将原始光谱通过一阶导、二阶导、标准正态法等方法进行数据预处理,并用灰色马尔科夫、logistics、偏最小二乘方法分别建立光谱数据与指标pH、挥发性盐基氮(TVB-N)之间的预测模型与分级模型。结果表明,pH与TVB-N含量最优预测模型均选择二阶导-偏最小二乘方法,相关系数分别为0.9973和0.9835,均方根误差分别为0.0383和1.5997。所建立模型的精确度符合检测要求,为冷鲜羊肉的分析测定提供一种快速方便的检测方法。

The multispectral(320-900 nm)imaging technology was used to establish a predictive model based on multispectral data for cold fresh lamb to realize fast non-destructive testing of cold fresh lamb.Scan cold fresh lamb with a spectrometer and the original spectrum was preprocessed by the first derivative,second derivative,standard normal method and other methods.The gray Markov,logistics and partial least squares methods were used to establish the prediction model and classification model between the spectral data and the indicators pH and volatile basic nitrogen(TVB-N)content.The results showed that the optimal prediction models for pH and TVB-N both used the second-order derivative-partial least square method.The correlation coefficients were 0.9973 and 0.9835,respectively,and the root mean square errors were 0.0383 and 1.5997.The accuracy of the established model met the detection requirements.A fast and convenient detection method for the analysis and determination of cold fresh lamb was provided.