基于太赫兹光谱和支持向量机快速检测棉花种子-《集美大学学报（自然科学版）》

文章信息/Info

Title:: Rapid Detection of Cotton Seed Based on THz Spectroscopy Combined with SVM

Author(s):: DU Yong1; LIU Jian-jun2; (1.School of Information Engineering,Jimei University,Xiamen 361021,China;2.School of Electronic Engineering,Jiujiang University,Jiujiang 330013，China)

摘要:: 鉴于目前对农产品品种的检测大多是基于可见光/近红外光谱的，提出了一种基于太赫兹光谱和支持向量机快速检测棉花种子的方法.为实现棉花种子的分类识别，在频率0.2~1.2 THz范围内采集2种最新转基因及2种非转基因棉花种子，总计40个样本的太赫兹光谱，用遗传算法优化的支持向量机建立识别模型，对不同品种的棉花种子进行识别.实验结果表明，该方法对不同品种的棉花种子综合识别率达到93.75%，由此，太赫兹光谱结合支持向量机的检测方法可为不同品种的生物辨别提供一种精确、快速、简便的检测方法.

Abstract:: At present,the detection of agricultural products is mostly based on visible/near infrared spectroscopy.In view of this,a fast and non-destructive detection method of cotton seeds based on terahertz spectroscopy combined with Support Vector Machine（SVM) was proposed.For the classification and recognition of different varieties cotton seeds,the terahertz spectra of two kinds of transgenic and two kinds of non-transgenic cotton seeds containing 40 samples in total were collected in the frequency range of 0.2~1.2 THz,using the Genetic Algorithm（GA)to optimized support vector machine.A recognition model to recognize different varieties of cotton seeds was established.The experimental results showed that the recognition rate of cotton seeds reached 93.75%.Therefore,the terahertz spectroscopy combined with support vector machine may provide an accurate,fast and simple method for the detection of different varieties of organisms.