计算机视觉用于猪肉新鲜度检测的颜色特征优化选取  

Optimized the color parameters in prediction of pork freshness by computer vision

作　　者：潘婧[1,2] 钱建平[2] 刘寿春[2] 韩帅[2] 

机构地区：[1]上海海洋大学信息学院,上海201306 [2]国家农业信息化工程技术研究中心,北京100097

出　　处：《食品与发酵工业》2016年第6期153-158,共6页Food and Fermentation Industries

基　　金："十二五"国家科技支撑计划资助项目(No.2013BAD19B04)

摘　　要：在以计算机视觉为基础,并利用神经网络预测猪肉通脊新鲜度时,选择合适的颜色特征参数和神经网络模型是提高其预测准确性的关键之一。文中提出了一种猪肉新鲜度等级预测时颜色特征参数和神经网络优化选取的方法,利用图像处理的方法提取通脊表面的颜色特征参数,组合成RGB-HIS、RGB-L*a*b*、rgb-HIS、rgb-L*a*b*及HIS-L*a*b*五类特征参数组合,并利用BP(back propagation,BP)和SVM(support vector machine,SVM)神经网络构造各类新鲜度等级预测模型。结果表明:SVM和BP的平均预测准确率分别为91.11%和84.44%,且rgb-HIS特征参数组合的BP与SVM预测模型的预测准确率最高,分别为88.89%和95.56%。因此,提取通脊表面r、g、b、H、I、S均值作为颜色特征向量,且选择SVM神经网络来构造新鲜度预测模型可显著提高预测结果的准确性。It is very important to choose the appropriate color feature parameters and the neural network model when using neural network to predict the freshness of pork tenderloin by computer vision. The method of selecting optimal color feature parameters and neural network was discussed. First,color characteristics parameters were extracted by imagine and formed 5 kinds color combinations such as RGB-HIS,RGB-L＊a＊b＊,rgb-HIS,rgb-L＊a＊b＊and HISL＊a＊b＊. Then the BP and SVM neural network were used to construct prediction model of all kinds of combinations.The result showed that the average prediction accuracy of SVM and BP was 91. 11% and 84. 44% respectively. RgbHIS and BP and SVM model had the highest prediction accuracy of 88. 89% and 95. 56% respectively. Therefore,extracting r,g,b,H,I,S from pork tenderloin as the color feature vectors and combined SVM neural network to construct the freshness prediction model can significantly improve the prediction accuracy.

关 键 词：猪肉 新鲜度 颜色特征 BP SVM 预测模型 

分 类 号：TS251.7[轻工技术与工程—农产品加工及贮藏工程]