一种基于高光谱图像的熟牛肉TVB-N含量预测方法  

A method for predicting TVB-N content of cooked beef based on hyperspectral image

作　　者：田卫新[1] 何丹丹[1] 杨东[2] 陆安祥[1,2] 

机构地区：[1]三峡大学计算机与信息学院,湖北宜昌443002 [2]北京市农林科学院北京农业质量标准与检测技术研究中心,北京100097

出　　处：《食品与机械》2016年第12期70-74,共5页Food and Machinery

基　　金：国家科技支撑项目(编号:2014BAD04B05-2)

摘　　要：传统肉制品新鲜度检测方法具有耗时费力、效率低、有损等缺陷,提出利用高光谱成像(HSI)技术预测熟牛肉新鲜度指标挥发性盐基氮(TVB-N)含量。首先通过HSI系统获取熟牛肉样本的高光谱数据,并进行黑白校正。进而采用移动平均平滑和多元散射校正对高光谱数据进行预处理。最后采用支持向量回归(SVR)方法分别建立基于全光谱特征、单一光谱特征、单一纹理特征、主成分分析(PCA)融合特征对TVB-N含量的预测模型。结果显示,使用PCA融合特征的SVR模型,对新鲜度的关键指标TVB-N含量的平均预测准确度(APA)可达到85.13%,表明高光谱成像技术与信息融合技术相结合能够提升模型准确度。Based on the shortcomings of the traditional detection methods for meat freshness,such as time-consuming,laborious,low efficiency, loss and other defects, and put forward using hyperspectral imaging（HSI）technology to predict cooked beef freshness index of volatile basic nitrogen（TVB-N）content.Firstly,the hyperspectral data of cooked beef samples were obtained by HSI system,and the black and white correction was carried out.And then,the hyperspectral data was preprocessed using the moving average smoothing and the multiple scattering corrections.Finally,the support vector regression（SVR）method was used to establish the prediction model of TVB-N content based on the whole spectral feature,single spectral feature,single texture feature and PCA fusion feature.The experimental results showed that the Average Predicting Accuracy（APA）for the TVB-N content index of freshness could reach85.13% by SVR model with PCA fusion feature,also showed that hyperspectral imaging technology combined with information fusion technology could improve the prediction accuracy of the model.

关 键 词：高光谱成像 TVB-N含量 特征融合 支持向量回归 熟牛肉 

分 类 号：TS251.7[轻工技术与工程—农产品加工及贮藏工程] TP391.41[轻工技术与工程—食品科学与工程]