Inversion Of Wheat Chlorophyll Content And Yield Based On Unmanned Aerial Vehicle Images

The Jianghuai region is an advantageous area for the growth of special wheat in China.In recent years,the growth of wheat in this region played an important role in the development of China's economy due to the huge market demand.Therefore,it is important to monitor the growth and forecasting yield of wheat.The traditional manual sampling method was destructive,which has the disadvantages including time-consuming and high-cost.Common-used satellite remote sensing could not maximize the advantages of large-scale monitoring due to the rainy climate in the Jianghuai region.It is urgent to introduce the popular UAV(Unmanned Aerial Vehicle)remote sensing technology.UAV remote sensing could monitor the grovwth and predict yield of wheat with the advantages of rapidness,accuracy and low-cost.In this thesis,two small UAVs were separately equipped with visible light and multi-spectral cameras to obtain field images in different growth stages of wheat from Shucheng and Baihu area.Then,the inversion models of the SPAD and yield of wheat were established based on vegetation indices by calculating images.Therefore,the growth could be monitored and the yield could be forecasted effectively.The study could provide an important reference to formulate suitable production and management methods for improving the quality and yield of wheat.The main contents of this thesis are as follows:(1)The "Feature Parameter Extraction of UAV Image and Wheat Chlorophyll Content and Yield Inversion System" was established on Matlab2016b.The system realizes RIO cutting,color space transformation,vegetation index calculation,measured value and vegetation index correlation analysis modeling and provides accuracy evaluation for the pre-processed UAV image,which provides the subsequent wheat chlorophyll content inversion test.Technical and data support.(2)UAV(DJI Phantom 3 Advanced)equipped with the visible light and ADC-lite multispectral camera was used to obtain the wheat images in flowering early and late stage in field(Shucheng County Agricultural Science Institute).respectively.Meanwhile,the SPAD 502 plus was employed to measure the chlorophyll content(SPAD)of wheat.Nine visible light vegetation indices and four multi-spectral vegetation indices of 30 cells in pre-processed visible and multi-spectral images were calculated,respectively.Then,the correlation coefficient between vegetation indices and the measured chlorophyll content of wheat was calculated and analyzed.Next,the two vegetation indices with the highest correlation of two cameras were selected to build the chlorophyll inversion model of wheat in flowering early and late stage.Correlation analysis showed that the visible vegetation index ExG(Excess Green)had the highest correlation with the measured chlorophyll content of wheat,and the coefficient of determination R2 was 0.659.The correlation between the NDVI(Normalized Difference Vegetation Index)and the measured chlorophyll content of wheat was the highest,and R2 was 0.692.The results indicated that the inversion model of NDVI-SPAD based on multi-spectral camera(flowering early stage model:R2=0.71 7,flowering late stage model:R2=0.613)was better than the ExG-SPAD model based on visible light camera(flowering early stage model:R2=0.687,flowering late stage model:R2=0.595).The results also showed that the model of early stage obtained the more accurate monitor than that of late stage.Which provided UAV remote sensing can effectively monitor crop growth in the Jianghuai region.(3)UAV(DJI Phantom 4 pro)equipped with a high-definition digital camera was applied to acquire wheat images in filling middle stage of 10 varieties wheat in Baihu Farm.Meanwhile,the chlorophyll content and yield were measured.A total of 12 color features were extracted and 8 visible light vegetation indices were calculated from the pre-processed high-resolution images(30 cells),respectively.Then,the correlation between calculated features and measured chlorophyll content,yield of wheat was explored and analyzed.The six characteristics with the highest correlation were selected as input variables.And the partial least square regression(PLSR)was used to construct the inversion model of chlorophyll content in filling middle stage and the yield of wheat.The results showed the correlation between COM(Combination Index),COM2(Combination Index 2),VEG(Vegetation Index),ExR(Excess Red),GRRI(Green-Red Ratio Index)and the measured chlorophyll content of wheat were the highest.The inversion model of chlorophyll content of wheat in filling middle stage based on PLSR was established using above characteristics,and obtained good fitting result with R2=0.550 and RMSE=1.899.Moreover,the correlation between R,NHLVI(Normalized Hue and Lightness Vegetation Index),YCbCr-Y,ExG,VDVI(Visible-band Difference Vegetation Index),ExR and the measured yield of wheat were the highest.The inversion model of the wheat yield based on PLSR was established using above characteristics,and also obtained relatively good fitting result with R2=0.831 and RMSE=0.132.The above research results could provide technique support for monitoring the growth and predicting the yield of special wheat in the Jianghuai region,which could also promote the development of precision agriculture and ensure security of food.