Abstract: The seedling emergence rate is the prerequisite and guarantee for high yield of spring-sown maize in Northwest China. In view of the phenomenon of heavy workload, time-consuming, labor-consuming and error leakage by manual statistics in the process of large-scale cultivation of maize in Ningxia, different nitrogen treatment experiments were designed, which used a drone equipped with a digital camera to obtain HD images of maize seedlings. The ORB algorithm in MATLAB software and distance-weighted fusion algorithm were used to synthesize unmanned aerial vehicle images. The image outlines of maize seedlings were obtained through depth optimization techniques such as binarization and corrosion expansion, and then automatically planned routes and calculated using MATLAB eight-bit connected domain and ARCMAP 10.3 calculation method. The number of maize emergence was routed and calculated. Based on the field artificial survey data, a linear regression analysis method was introduced to establish a linear relationship model between manual counting and the number of maize seedlings obtained by drones. The results showed that the determination coefficient of the linear regression relationship model, root mean square error and normalized root mean square error were 0.895, 4.359 and 2.436%, respectively. Therefore, based on the low-altitude drone platform, the rapid acquisition of the number of emergence of maize in the field was a time-saving, efficient and accurate method for obtaining seedling rate. It could provide technical support for the accurate assessment of the subsequent high yield of maize, and was of great significance for optimizing the precise planting technology of water and fertilizer integration of maize in Ningxia.

Key words: maize, emergence rate, drone, image synthesis, image recognition, image processing, regression analysis