Timely and accurate grasp of crop plant nitrogen content (PNC) information is helpful to monitor crop growth and realize the scientific management of farmland nitrogen fertilization. Based on this, taking unmanned aerial vehicle (UAV) as the platform to obtain digital images of potato budding, tuber formation, tuber growth, starch accumulation, and maturity period, and the PNC, plant height, and the three-dimensional coordinates of the ground control point (GCP) were measured. Secondly, the digital orthophoto map (DOM) and digital surface model (DSM) of the test area were generated by combining the digital images of UAV in each growth period with GCP. Then, the correlation analysis between the Hdsm and the constructed image variables of each growth period with the PNC measured on the ground were carried out, and the image variables with good correlation were selected as the input parameter of the potato PNC estimation models with the Hdsm. Finally, based on the image variables and image variables combined with Hdsm, three methods of multiple linear regression (MLR), error back propagation (BP) neural network, and Lasso regression were used to construct the PNC estimation models of potato at each growth stage. The results showed that the Hdsm extracted based on DSM had a high degree of fit with the measured H(R2 was 0.860, RMSE was 2.663cm, and NRMSE was 10.234%). Adding Hdsm in each growth period can improve the accuracy and stability of estimating potato PNC. The effect of PNC estimation model constructed by MLR method in each growth period was better than that of BP neural network and Lasso regression. Therefore, the research result can provide a technical reference for the efficient and non-destructive monitoring of potato PNC status.