Multi-temporal hyperspectral image change detection method based on multi-scale feature fusion

Abstract

The application discloses a multi-temporal hyperspectral image change detection method based on multi-scale feature fusion, adopts a double-branch network structure, and is used for extracting features of different time phases. Meanwhile, in order to accurately capture complex spectral features, the dynamic convolution module is used for considering the features in multiple scales, the channel attention mechanism is introduced, the adaptive adjustment of different band features is realized, and the perception and capture ability of the model for complex spectral features in hyperspectral data is improved. The multi-scale feature fusion module runs on deep features, through multi-scale feature extraction and fusion, the perception ability of the model for local information is enhanced. Through experiments on three real-world hyperspectral image change detection data sets, the results show that, compared with current advanced change detection methods, the application has significant superiority in effectiveness and performance.

Images