Evaluation Of Seed Emergence Uniformity Of Wheat Based On UAV Image

Nowadays,in order to make the agricultural production more efficient and intelligent under the background of Internet plus,information technology has to be integrated into the development of modern agricultural,which is a trend of modern agricultural development in China and around the world.Under the background of the development of Internet plus agricultural,the evaluation of seed emergence uniformity of wheat based on UAV image was studied in this paper.The purpose is to prove a new intelligent method for detecting the emergence of wheat seedlings.The uniformity of wheat seedling emergence is an important part of the diagnosis of wheat field,which would greatly influence the growth and development of wheat.In this study,the field images of wheat were obtained by using UAV,which were preprocessed by image processing technology,and the estimation model of wheat seedling umber was constructed by multiple stepwise regression method.The method of reflecting the uniformity of wheat seedling emergence was based on the image processing technology and the above model for estimating the number of seedlings.Finally,the uniformity of wheat emergence was evaluated with the output data.The main conclusions are as follows:(1)The relationship between the number of wheat seedlings and coverage at the 1-3 leaf stage were analyzed under the conditions of sowing method and density test in this study.Models for estimating the number of wheat seedlings in different periods were established by using coverage.The results show that the R2 value of the estimation model was highest at the lowest density in the 1 leaf stage.The R2 value showed a downward trend with the increase of the density or leaf age.The correlation between cover and wheat seedlings was poor in the late period of fertility and high density,although they were highly relevant in the early period.(2)Multiple indicators including coverage,number of corner points,leaf age and variety were used to construct the estimation model in order to build the best estimate model in this study.The results show that the R2 values were more than 0.95 in the process of modeling and the RMSE values were also in a smaller range.The model was verified by comparing the estimated values and the measured values.Although the model had some differences in different density and different leaf age,the R2 values of the model were above 0.95,and the RMSE values were smaller under low density conditions and the R2 values were above 0.85 under high density conditions.The accuracy of the model was highest in one leaf stage and lowest in the 3 leaf stage,but the average value of R2 was above 0.9.(3)Images of wheat seedling stage under different density and different sowing methods were obtained by UAV,and each image was divided into several parts.The number of seedlings each part was acquired quickly through the above seedling number estimation model.Seedling uniformity was reflected by calculating the reciprocal of CV(coefficient of variation)of the number of seedlings per aliquot.We compared the seedling uniformity obtained by artificial investigation and image analysis technology and found that the average error of the variation values calculated by image processing is less than 10%.The value of RMSE is smaller and the value of R2 is above 0.7.The above results showed that the technology of UAV image processing proposed in this study can accurately reflect the uniformity of wheat seedling emergence.(4)The relationships between emergence uniformity and yield and its components were analyzed under the sowing of three density conditions.When the evenness of emergence was better,the change of yield was not obvious.With the decrease of uniformity of emergence,the yield of wheat decreased.Spike numbers and per spike grains were significantly positive correlated with seedling uniformity.As the rise of the seedling uniformity,the 1000 grain weight had a rising trend,but the trend was not obvious.Under the condition of sowing 240×104·ha-1 density,seedling uniformity of 20 different treatments were processed by cluster analysis.When the distance between classes was about 7,it could be divided into 3 categories:6470-6837kg·ha-1?7060-7314kg·ha-1?5598-6324kg·ha-1.When the seedling uniformity is higher than 3.01,the yield would be less affected by the uniformity of seedling emergence.When the variation coefficient of emergence uniformity was in the range of 2.33-3.01,the yield would be more susceptible to the uniformity of emergence.When the coefficient of variation of the emergence uniformity was less than 2.33,the yield would be affected more by the uniformity of seedling emergence.