Inversion of Nitrogen Content in Wheat Based on Multi-Spectral Images from UAV

The nitrogen content of wheat is an indispensable influencing factor during the growth and development of wheat, and it has a direct impact on the growth and development of wheat. To improve the accuracy and comprehensiveness of monitoring the nitrogen content of wheat, this study selected 8 typical areas in the main wheat-growing areas of Feicheng County, Shandong Province as experimental sites. Based on the field rapid measurement data of wheat nitrogen content, combined with the multi-spectral images of unmanned aerial vehicles (UAVs), the correlation between the field rapid measurement of wheat nitrogen content and vegetation indices was analyzed, and the vegetation index with an absolute value of the correlation coefficient greater than 0.7 was selected as the input variable. Three models, namely Multiple Linear Regression (MLR), Back Propagation-Artificial Neural Networks (BP-ANN), and Particle Swarm Optimization-Back Propagation Neural Network (PSO-BP), were constructed respectively, and the accuracy of each model was evaluated. The results show that among the three models, the N-PSO-BP model is the optimal inversion model. Its R2 in the validation set is 0.898, and compared with N-MLR and N-BP-ANN, the R2 in the validation set is increased by 14.1% and 5.3% respectively, and the RMSE is reduced by 31.7% and 20.4% respectively. This indicates that the N-PSO-BP model constructed based on multi-spectral images of UAVs can better reflect the actual situation of wheat nitrogen content in the aerial survey area and is suitable for monitoring the nitrogen content of wheat using multi-spectral UAVs. This study can provide data support for monitoring the growth of wheat and provide decision-making basis for regional agricultural production and precision agriculture management.