Extraction And Growth Analysis Of Maize Plant Numbers At Different Growth Stages Based On UAV Remote Sensing

Maize is one of the most important food and commercial crops in our country.During the maize growing period,accurate and rapid acquisition of growth information such as the number of maize plants and growth is of great significance for yield estimation,quality assessment and field management.In the fine field management of maize,satellite images are difficult to monitor crop growth information in time due to the limitations of long revisit period,low spatial resolution and other external factors.Therefore,there is an urgent need for a fast,non-destructive,high-precision means and method that can monitor crop growth over a large area.UAV hyperspectral remote sensing has the characteristics of high spatial resolution and spectral resolution,flexible and rapid acquisition of images,low cost and simple operation,and has become an important data source for field management agricultural remote sensing.In this study,UAV flight experiments were carried out in Shuguang Village,Jinhe Town,Saihan District,Hohhot City,Inner Mongolia,and drone images were collected in seven growth stages,including seedling stage,three-leaf stage,jointing stage,small trumpet stage,large trumpet stage,male pumping stage and filament drawing stage,and data such as the number of maize plants,plant height,chlorophyll content and leaf spectrum were collected.The number of maize plants at different growing stages was extracted by threshold segmentation and verified with field measurements.The UAV point cloud data was used to construct DSM at different growth stages,and the maize plant height was extracted.Through the change of chlorophyll content in maize leaves,the variation of chlorophyll content in different growth stages with maize growth was analyzed,and its correlation with the spectral index was analyzed,and the model for estimating chlorophyll content was constructed by RF(random forest),SVR(support vector regression)and BP(neural network),and the model that best reflected the growth index was selected.In addition,based on segmented images combined with image classification method,the coverage of maize at different growth stages was calculated,and the change of coverage was analyzed.The specific results are as follows:(1)The extraction results of the number of corn plants showed that the extraction accuracy at the emergence stage reached 0.98,and the R~2value reached 0.81.(2)In terms of growth analysis,it can be seen from the changes of vegetation indices in different growth periods that the trend of 8 planting index indexes,including NDVI,GNDVI,DVI,EVI,SAVI,RVI,PRI,CCCI,etc.,is consistent,and it is increased from the emergence stage to the trumpet stage,while the change from the large stage to the drawing stage is different.Among them,NDVI,GNDVI,EVI,SAVI,RVI decreased during the male pumping period,but DVI,PRI,and CCCI were on the rise.The trend of OSAVI declines with the growth of corn,with OSAVI lowest in the drawing period.(3)From the changes of chlorophyll content in different growth stages,it can be seen that the chlorophyll content of maize increases first and then decreases with the growth period,and has the highest leaf chlorophyll content in the big trumpet stage.Among the growth period estimation models,the SVR model in the drawing period had the best estimation effect,and its accuracy and stability were good.(4)From the change of coverage in different growth periods,it can be seen that from the emergence stage to the filament drawing stage,the overall coverage of maize in the study area shows a gradual increase trend with the growth of maize,and the trend from low to high during the entire growth period.(5)From the change of plant height in different growth periods,it can be seen that during the entire growth period,the trend of corn plant height changes from slow to rapid growth and then slow trend.