Study On Rapid And Non-destructive Measurement Of Rice Seed Vigor Using Spectrum

Rice is the main food on which we live,and seed vigor level is an important factor to judge seed quality.Traditional methods of seed vigor detection are too complicated,so it is necessary to explore a new effective,rapid and non-destructive method to identify seed vigor.Near infrared(NIR)and hyperspectral(HYPERspectral)techniques are relatively mature emerging technologies in recent years,which can effectively identify seed vigor.In this study,n IRS and hyperspectral techniques were used to determine the viability of different rice varieties under different storage conditions,combined with linear regression and random forest modeling analysis.Based on the actual germination or not,combined with the determination of physiological and biochemical indicators,and then fusion of near-infrared spectroscopy(near-infrared band 900nm-1700 nm,resolution of 5 nm)and visible light(visible band 400nm-1000 nm,resolution of 3 nm)data for prediction analysis.In order to test the model to identify the seed germination accuracy.The main conclusions of this study are as follows.1.Seed germination index results of different rice varieties under different storage conditionsAccording to the results of standard germination test,the vigor of rice seeds stored in cold storage was higher than that stored in natural storage.The maximum difference of germination rate was 93%.2.Results of physiological and biochemical indexes of seeds of different rice varieties under different storage conditionsGermination indexes(germination rate,germination potential,vigor index,simple vigor index and germination index)of seeds of different rice varieties were positively correlated with physiological and biochemical indexes(POD enzyme activity,CAT enzyme activity,protein content and starch content),and the higher the germination rate was,the higher the physiological and biochemical indexes were.These physiological indexes are generally higher in cold storage than in natural storage.The longer the storage time,the lower the CAT enzyme activity and POD enzyme activity,the lower the protein content and starch content.3.Accuracy of determination of rice seed vigor by near infrared technologyUnder different wavelength conditions,the curve of average reflectance of different varieties of grain is basically the same,all of which show wavy change.However,the average reflectance of seeds of different rice varieties is different,and the spectral characteristics of seeds of the same rice variety are obviously different from that of brown rice.The average value of cross validation under natural storage condition was 75.33%,and that under cold storage condition was 72.08%.The accuracy of model discrimination based on random forest is 74.00% and 68.80%.With 533 groups was established under the condition of natural rice seed storage,on the basis of the related model to the actual measurement of long ping high-tech rice seed germination as the validation criteria,validated for long ping high-tech rice seed,the results show that the natural storage condition prediction accuracy is 50.40%,refrigerator storage under the condition of prediction accuracy is 53.30%.4.Accuracy of determination of rice seed vigor by hyperspectral techniqueThe hyperspectral curves of rice grains and brown rice of different seeds decreased first and then increased.Under the condition of different wavelength,the spectral characteristic values of the same variety of grain are different,and the spectral characteristics of the same variety of seed grain and brown rice are obviously different.The average value of cross validation under natural storage condition is 84.49%,and that under cold storage condition is 86.21%.The accuracy of model discrimination based on random forest is 71.40% and 75.30%.With 533 groups was established under the condition of natural rice seed storage,on the basis of the related model to the actual measurement of long ping high-tech rice seed germination as the validation criteria,to verify this long ping high-tech seeds,the results show that the natural storage condition prediction accuracy is 48.50%,refrigerator storage under the condition of prediction accuracy is 53.70%.In conclusion,the longer the seed storage time,the lower the germination index.Storage time and conditions will affect the changes of seed material.In this experiment,seed vigor of different rice varieties can be predicted basically.The prediction accuracy of the model based on the cold storage condition of rice seeds is higher than that of the natural storage condition,no matter in the near infrared spectrum or hyperspectrum.The reasons for this result need to be further explored.