The invention discloses a multi-feature fusion image classification method based on deep learning. The method specifically comprises the steps of data set division, data enhancement, classification network model construction, model initialization and model training optimization. And the data enhancement part is used for enhancing data characteristics by randomly carrying out operations such as horizontal overturning, vertical overturning, brightness modification and horizontal overturning according to probability on the picture. In the construction process of the classification network model,the features extracted for the first time are randomly covered and then extracted again, and then the features extracted for the two times are fused, so that the features are diversified, and the classification accuracy is improved. The system can be used for classifying eye malignant tumor images, positioning lesion areas in the images as feature areas, giving out probability values of lesion types and assisting film reading doctors in judgment.