基于高光谱信息的生鲜鸡肉离心损失率快速预测模型构建  

Model Establishment for Fast Predicting Centrifugal Loss Rate of Fresh Chicken Based on Hyperspectral Information

作　　者：何鸿举[1,2] 王洋洋 王魏 蒋圣启 朱亚东 马汉军[1,2] 陈复生[2,3] 王玉玲 朱明明 赵圣明[1] 潘润淑[1] HE Hong-ju;WANG Yang-yang;WANG Wei;JIANG Sheng-qi;ZHU Ya-dong;MA Han-jun;CHEN Fu-sheng;WANG Yu-ling;ZHU Ming-ming;ZHAO Sheng-ming;PAN Run-shu(School of Food Science,Henan Institute of Science and Technology,Xinxiang 453003,China;Postdoctoral Research and Development Base,Henan Institute of Science and Technology,Xinxiang 453003,China;College of Grain,Oil and Food,Henan University of Technology,Zhengzhou 450001,China;School of Life Science and Technology,Henan Institute of Science and Technology,Xinxiang 453003,China)

机构地区：[1]河南科技学院食品学院,河南新乡453003 [2]河南科技学院博士后研发基地,河南新乡453003 [3]河南工业大学粮油食品学院,河南郑州450001 [4]河南科技学院生命科技学院,河南新乡453003

出　　处：《食品工业科技》2020年第9期238-243,共6页Science and Technology of Food Industry

基　　金：河南省重大科技专项项目(161100110600);中国博士后科学基金(2018M632767);河南省科技攻关项目(182102310060);河南省青年人才托举工程项目(2018HYTP008);河南省博士后科研项目(001801021);河南科技学院高层次人才引进项目(2015015);河南科技学院重大科研培育项目(2016ZD03)。

摘　　要：本文旨在挖掘900~1700 nm波长范围内的高光谱信息构建生鲜鸡肉离心损失率的快速预测模型。通过采集生鲜鸡肉样品的高光谱图像,并提取图像感兴趣区域的光谱信息,经基线校正(Baseline Correction,BC)、高斯滤波平滑(Gaussian Filter Smoothing,GFS)、多元散射校正(Multiplicative Scatter Correction,MSC)、移动平均值平滑(Moving Average Smoothing,MAS)、中值滤波平滑(Median Filtering Smoothing,MFS)5种光谱预处理后,建立全波段偏最小二乘(Partial Least Squares,PLS)回归模型,并利用回归系数法(Regression Coefficient,RC)、连续投影算法(Successive Projections Algorithm,SPA)和逐步回归法(Stepwise)筛选特征波长,优化全波段模型。结果显示,基于Stepwise法从原始光谱中筛选的16个最优波长(900.6、915.4、1024.0、1089.8、1111.2、1155.6、1165.5、1288.9、1305.4、1433.9、1442.1、1486.7、1493.3、1541.1、1690.1和1693.4 nm)构建的PLS模型预测效果较好,其中,rC为0.94,RMSEC(Root Mean Square Error of Calibration)为1.43%,rP为0.94,RMSEP(Root Mean Square Error of Prediction)为1.60%。本文表明,基于高光谱信息构建的PLS模型可快速预测生鲜鸡肉离心损失率。The present study aimed to establish a rapid model for predicting centrifugal loss rate of fresh chicken by mining hyperspectral information in the wavelength range of 900~1700 nm.Hyperspectral images of chicken samples were acquired and the spectral information within the region of interest of images were extracted.Partial least squares(PLS)models were established based on the full range wavelengths pretreated by baseline correction(BC),Gaussian filter smoothing(GFS),multiplicative scatter correction(MSC),moving average smoothing(MAS)and median filtering smoothing(MFS),respectively.Through regression coefficient(RC),successive projections algorithm(SPA)and stepwise algorithms,optimal wavelengths were respectively selected to optimize the full wavelength PLS model.The results showed that the PLS model based on the 16 optimal wavelengths(900.6,915.4,1024.0,1089.8,1111.2,1155.6,1165.5,1288.9,1305.4,1433.9,1442.1,1486.7,1493.3,1541.1,1690.1 and 1693.4 nm)selected from raw spectra by stepwise method had better performance,with rC of 0.94,root mean square error of calibration(RMSEC)of 1.43%,rP of 0.94 and root mean square error of prediction(RMSEP)of 1.60%.The experiment concluded that PLS model based on hyperspectral information could be used for the rapid prediction of centrifugal loss rate in raw fresh chicken.

关 键 词：高光谱 生鲜鸡肉 逐步回归法 偏最小二乘 离心损失率 

分 类 号：TS251.7[轻工技术与工程—农产品加工及贮藏工程]