| With the development of agricultural production technology,the cultivation methods of crops tend to be perfect.The effect of seed quality on increasing agricultural income and increasing agricultural production is becoming increasingly important.As an important grain,feed and industrial crop in China,corn plays an important role in agricultural production and the national economy.The quality of maize seeds is the key factor affecting the yield and quality of maize.The detection of seed quality is an important part of agricultural production,including variety identification and vigor identification.Traditional detection methods take a long time,slow detection,and damage to the sample,which seriously restricts the development of the corn seed industry.The subject uses terahertz spectroscopy technology and terahertz imaging technology combined with neural network,pattern recognition and chemometric methods to determine the variety and vigor of corn seeds.Perform fast,non-destructive testing to monitor seed quality.The main contents of this thesis are as follows:Firstly,the terahertz time-domain spectroscopy technique combined with neural network learning vector quantization method was used to establish a corn seed cultivar identification model.The results showed that the model prediction accuracy rate was over 80%,which could effectively identify corn varieties.Therefore,this method has certain reference to the rapid identification of crop varieties.Secondly,the terahertz time-domain spectroscopy technique combined with the moving window correlation coefficient method is used to analyze the mechanism of the change of vitality during the aging process of corn seed and extract the characteristic band.The results show that the change of lipid during the aging process of corn seed causes the change of embryo.Changes in starch cause changes in the endosperm.Therefore,changes in seed and endosperm can be studied by monitoring changes in lipids and starch in corn seeds to study changes in corn vigor.Thirdly,the terahertz reflection imaging technology combined with image reconstruction technology is used to extract the spectral information of different tissues of seeds,and the support vector machine discriminant model of seed germination rate is established.The results show that the recognition rate of seed qualitative analysis model can reach more than 85%.Therefore,terahertz imaging technology can achieve the identification of seed vigor levels. |
