食品科学 ›› 2008, Vol. 29 ›› Issue (6): 41-44.

• 基础研究 • 上一篇    下一篇

Fisher线性判别法在干酪分类鉴别中的应用

 贾宗艳, 田立军, 郑丽敏, 朱虹, 芦筱菲, 任发政   

  1. 中国农业大学信息与电气工程学院; 中国农业大学信息与电气工程学院
  • 出版日期:2008-06-15 发布日期:2011-08-26

Application of Fisher Linear Discriminant Method in Cheese Flavors Differentiation

 JIA  Zong-Yan, TIAN  Li-Jun, ZHENG  Li-Min, ZHU  Hong, LU  Xiao-Fei, REN  Fa-Zheng   

  1. College of Information and Electrical Engineering, China Agricultural University
  • Online:2008-06-15 Published:2011-08-26

摘要: 基于气敏传感器阵列和模式识别技术,本研究提出了一种依据干酪的气味对干酪进行分类的方法。对国内光明、三元、台湾Anchor、法国百吉福、欧洲多美鲜5种不同品牌的奶酪气味进行了采样,并从每个传感器的响应曲线上提取了6个特征值。对利用主成分分析的方法对特征向量进行了降维,并建立了特征库。利用Fisher线性判别法建立了区分干酪的两种线性分类器,并重新采样对两种分类器进行了测试比较,其中一种方法的识别的正确率仅为50%,另一种达到了100%。

关键词: 干酪, 电子鼻, Fisher, 模式识别

Abstract: Based on the sensory array and pattern recognition, a new method to differentiate the cheese by their different flavors was put forward. Five brands of different cheese flavors,such as Guangming, Sanyuan, Anchor, Pikifou and Suki, were sampled,and six features on the response curve filtered of every gas sensor were picked up. There were thirty features for every sample. And then the features were extracted with the principal component analysis (PCA). The feature database was built up. Two linear classifiers were established with Fisher Linear Discriminant Method. Being tested, the accuracy of discrimination with one method is 50%, and that of another one is 100%.

Key words: cheese, electronic nose, Fisher, pattern recognition