Study On Evaluation Method Of Maize Seedling Emergence Quality Based On UAV Image

Objective:A fast and accurate corn seedling number monitoring algorithm was constructed based on the image data of corn seedling stage acquired by visible light imaging equipment carried by UAV.The influence of different factors such as maize leaf age and planting density on the monitoring accuracy rate of maize seedling number was analyzed,and the applicable monitoring method was established.The evaluation method of maize emergence quality based on image monitoring was constructed to provide data reference for post-seedling field management and decision making.Methods:The experiment was carried out in Qianjiadian Town,Tongliao,Inner Mongolia.Djiang six-rotor UAV equipped with hd digital camera was used to obtain hd RGB visible image data of corn seedling stage(2 leaves,3 leaves,4 leaves and 6 leaves).Open CV software library in Python was used to perform threshold segmentation to segment vegetation and soil background by ultra-green index(EGI,(2G-R-B)/G).Combined with image contrast enhancement technology,morphological operation,contour detection and recognition,the algorithm was constructed to obtain the number of corn seedling quickly and accurately.Based on the emergence number detection algorithm,the accuracy of the algorithm was analyzed for maize population with different planting densities and different leaf ages.The accuracy of corn emergence quantity detection and the accuracy of maize emergence quality and uniformity evaluation method were analyzed for the common seeding machine field driven by navigation-assisted driving seeder and machine operators of different technical levels.Results:(1)The emergence number detection algorithm based on image processing acquired by UAV platform can better identify maize seedlings at specific leaf age and accurately obtain the number.(2)There were differences in the monitoring accuracy at different leaf ages.Under different planting densities,the detection error decreased first and then increased with the growth of maize seedlings,and the error was the lowest at 4-leaf stage,followed by 3-leaf stage.The monitoring effect at each density was the best at the 4-leaf stage,with r RMSE between 1%-3%and R~2 of image counting model higher than 0.9.The monitoring effect was better at the 3-leaf stage,and the r RMSE at each density was between 4%-6%,and the R~2 of image counting model was higher than 0.8.The monitoring error was higher in other periods,and the r RMSE was lower than 5%at low density(30,000 plants/ha and 45,000 plants/ha)at six leaf stage,and above 10%at other periods and densities.(3)Based on the number of emergence algorithm,establishes the evaluation index,while the quality index and seedling emergence uniformity build emergence quality evaluation method,and through three different ways of planting and seeding level of farmers to verification,evaluation result is consistent with the farmers to actual plant level,build the emergence of quality assessment methods,real time,objective and accurate evaluation of the quality,It can improve the accuracy of seedling quality assessment,provide basis for post-seedling management,realize the fixed remuneration according to seedling,and play a role in supervising the operation of hired machine operators and improving the quality of seeding.Conclusion:The algorithm proposed in this experiment to obtain the number of seedlings by UAV visible light image is feasible.When using UAV to monitor the number of seedlings,it is advisable to carry out at3-4 leafing stage,and the best time is at 4 leafing stage.Combined with UAV image processing technology and field experiment,a set of objective and accurate evaluation method of seedling emergence quality and uniformity of seedling emergence rate was proposed.