Mining land use classification method and system based on adversarial fusion features and transformers

Abstract

The application discloses a mine land use classification method and system based on an adversarial fusion feature and a transformer, and comprises the following steps: applying multispectral image (MSI) and synthetic aperture radar image (SAR) data to mine land use classification. First, data is preprocessed and feature extraction is performed. Then, the features are fused by using an adversarial learning mechanism, and GLCM texture features are introduced to enhance the classification effect. Next, the processing capacity of the model for global and local information is improved by using a transformer coding network. Finally, the classification model is trained and optimized by using a cross-entropy loss function. The application has the advantages that the classification accuracy and robustness are significantly improved.

Images