In order to accurately diagnose the potassium content status of apple leaves in each growth period, a diagnostic model for potassium deficiency in apple leaves based on shape and color combination feature is constructed. Firstly, a series of image preprocessing work such as image denoising, leaf segmentation, etc. are carried out on leaf image samples in each growth period. Secondly, 9 color characteristics and 10 shape characteristics of a leaf are extracted by digital image processing technology, and the data dimension reduction and optimization are carried out through linear discriminant analysis method to obtain the key shape and color combination feature factors of apple tree leaves in each growth period; then, the established LDA-SVM, LDA-RF and LAD-KNN models are compared with the accuracy of potassium deficiency diagnosis of apple leaves at different periods to obtain the best diagnostic model for each growth period. Finally, the best diagnostic model is used for field experiments, and the generalization ability and robustness of the model are verified by the results. The test results show that the diagnostic accuracy of the model reaches an average of more than 93.6% in the whole growth cycle, which can accurately diagnose the potassium content of apple leaves in each growth period, and provide methods and ideas for the intelligent management of orchards and supplementary fertilizer application information.