Study On Rice Quality Monitoring Method Based On High Resolution Data

In the new era,people pay more and more attention to the quality of crops.Crude protein and amylose content affect the nutrition and taste quality of rice,which are two important indexes of rice quality.Because of its fast,nondestructive and dynamic characteristics,remote sensing technology has shown strong vitality in the monitoring and testing of crop quality.Rice quality remote sensing monitoring started earlier in foreign countries,and in recent years,some progress has been made in China.In this paper,we studied the method based on high resolution remote sensing data(high resolution data)to monitor the crude protein and amylose content of rice,which provided the basis for the rapid and large-scale monitoring of rice quality.In this study,we selected the eastern rice field in eastern Tiaoxi,Deqing County,and carried out the experiment at the mature period of rice 2013-2014.The hyperspectral data of different forms of rice(ear of rice,paddy,rice grain,rice flour)were collected to monitor the quality of rice,and we studied the application of field high resolution satellite data,too.The main results as follows:In the indoor hyperspectral study,the data of rice were collected and we analyzed the correlations between six spectral varibles and the quality of rice.Single factor model,stepwise regression model,partial least squares model and support vector machine regression estimation model were adopted.It’s found that the paddy’s spectra was more more suitable for monitoring the amylose and crude protein content of rice.Compared different models,we found the partial least squares regression model established by the first derivative transformation of spectral data is the optimal estimation model of crude protein content.It’s the optimal model to estimate amylose content by using the reflectance values of paddy at 440 nm,680nm,550 nm,990nm and 1200 nm to establsih the support vector machine regression.In the study on the quality monitoring by using satellite data,we used HJ CCD and GF-1 WFV satellite data to analyze the quality indexes of rice.Firstly,the spatial distribution of rice was extracted by hierarchical classification method,which was composed of decision tree classification,visual interpretation and object-oriented classification.Then we analyzed the correlations between 14 kinds vegetation indexs and the quality indexes of rice,established models to estimated the indexes.Compared the result of 2014,it showed that HJ CCD data is more suitable for the estimation of crude protein content.This is because the range of crude protein content is small,less affected by spatial resolution,and the HJ CCD data was closer to the sampling date.For the estimation of amylose content with large content range,although the correlation between GF-1 WFV data and amylose content is better,its verification accuracy is not good.Finally,based on the optimal model established by the RDVI index of HJ CCD data from 2013 to 2014,the spatial distribution of crude protein content of rice in the study area was obtained.In this study,we established a monitoring model of crude protein content of rice based on high resolution satellite data and models of crude protein and amylose content monitoring based on hyperspectral data of indoor paddy.It was necessary to design scientific and rigorous experiments on other forms(ear of rice,paddy,rice flour)of the quality,and need to synchronize with the satellite transit canopy spectral observations.The use of satellite amylose content monitoring methods still need to be improved.These contents will be reflected in future research work,with a view to the use of satellite data to monitor the quality of rice to provide a reliable basis.