以市售新鲜冷藏(4℃)鸡胸肉为研究对象,采集鸡胸肉表面的高光谱(400~1 100 nm)图像信息,采用偏最小二乘回归(partial least square regression,PLSR)建立菌落总数预测模型,采用不同预处理方法提高模型的预测准确性和稳健性,实现快速无...以市售新鲜冷藏(4℃)鸡胸肉为研究对象,采集鸡胸肉表面的高光谱(400~1 100 nm)图像信息,采用偏最小二乘回归(partial least square regression,PLSR)建立菌落总数预测模型,采用不同预处理方法提高模型的预测准确性和稳健性,实现快速无损检测生鲜鸡胸菌落总数的目的。结果表明:标准变量变换(standard normalized variate,SNV)预处理后,模型性能最佳。模型的校正标准差(standard error of calibration,s_(EC))和验证标准差(standard error of prediction,s_(EP))分别为0.40和0.57,s_(EP)/s_(EC)为1.08,校正集相关系数(correlation coefficient of prediction,R_C)和验证集相关系数(correlation coeffic ient of prediction,R_P)分别为0.93和0.86;且应用最佳模型可有效预测样品菌落总数的分布地图。展开更多
文摘以市售新鲜冷藏(4℃)鸡胸肉为研究对象,采集鸡胸肉表面的高光谱(400~1 100 nm)图像信息,采用偏最小二乘回归(partial least square regression,PLSR)建立菌落总数预测模型,采用不同预处理方法提高模型的预测准确性和稳健性,实现快速无损检测生鲜鸡胸菌落总数的目的。结果表明:标准变量变换(standard normalized variate,SNV)预处理后,模型性能最佳。模型的校正标准差(standard error of calibration,s_(EC))和验证标准差(standard error of prediction,s_(EP))分别为0.40和0.57,s_(EP)/s_(EC)为1.08,校正集相关系数(correlation coefficient of prediction,R_C)和验证集相关系数(correlation coeffic ient of prediction,R_P)分别为0.93和0.86;且应用最佳模型可有效预测样品菌落总数的分布地图。
基金supported by the National Natural Science Foundation of China(52172171,52090022,11725210,and 91963115)the Natural Science Foundation of Hebei Province of China(E2022203109)Dr.Nie A appreciates the support of the Natural Science Foundation for Distinguished Young Scholars of Hebei Province(E2020203085)。
