Abstract: Aimed to accurately predict the nitrogen content of apple leaves in the fruit expansion stage, maturity stage, and harvest stage, a prediction model of apple leaves nitrogen content based on image processing was proposed. Firstly, the crown image and the new-shoot leaf image of the apple tree at different growth stages were collected by using UAV and digital camera in the visible spectrum. Then, the morphological characteristics of the new-shoot leaf image and the color characteristics of the apple tree crown were extracted by using digital image processing technology. Secondly, the nitrogen content of apple tree leaves was measured by the Kane determination method. Besides, the polynomial regression model, SVM model, and artificial neural network model have been constructed for the extracted image characteristic parameters and nitrogen content of apple tree leaves. According to the model verification results, the optimal model was determined to be an artificial neural network model. Finally, the performance of nitrogen content prediction models at three different stages of fruit expansion, maturity, and the harvest was tested to determine the accuracy of the prediction model. The test results showed that the root-means-square error（RMSE） of the three different periods of apple tree leaf nitrogen content prediction models are 0.039, 0.029, 0.037, and the mean-absolute errors（MAE） were 0.338, 0.403 and 0.412, and the mean-absolute-percentage errors（MAPE） were 0.582, 0.635 and 0.642, The predicted value of the model fits well with the actual value. respectively. This method can real-time monitor the growth and nitrogen status of apple trees, and provide scientific reference for the accurate fertilization management of orchards.