基于电子鼻的水稻品种鉴别研究

为了实现水稻品种的快速鉴别,避免水稻品种混杂,利用电子鼻对来自同一产地不同水稻品种进行测试,获取有效信息。对获取的信息提取平均微分值和面积斜率比两种特征。采用主成分分析、Fisher判别分析及BP神经网络3种模式识别方法进行水稻品种的判别,并对3种识别方法的结果进行比较分析。结果表明:不同种类的水稻品种可以被区分开来,但BP神经网络分类效果最好,Fisher判别分析效果次之,PCA分类效果最差。因此,结合合适的特征提取方法及模式识别方法,有可能实现一种基于电子鼻技术的对不同水稻品种鉴别的简单、有效的方法。

The research of rice varieties identification by electronic nose

In order to identify rapidly and avoid hybrids of different rice varieties,electronic nose is used to test rice varieties from different areas and obtain valid information.There are two types of features for obtained information to extract average differential value and area ratio of the slope.The different rice varieties are distinguished by principal component analysis(PCA),Fisher discriminant analysis(FDA)and BP neural network,and the results of the three identification methods are compared and analyzed.The results show that:different rice varieties can be distinguished,however,the discriminant result of BP neural network is the best,Fisher discriminant analysis follows and PCA analysis is the worst.Therefore,it can be an effective way to distinguish the different rice varieties by the electronic nose when the proper methods of feature extraction and pattern recognition are taken.