Full-Range Fusion Network with Local-Global Attention for Change Detection in Remote Sensing Images

Remote sensing image change detection (CD) is an important technology used to monitor changes in surface features and objects over time, widely applied in fields such as land use planning, environmental monitoring, and disaster management. Although traditional change detection methods are effective in certain scenarios, they are often limited by high sensitivity to noise and computational complexity, making them unsuitable for the rapidly evolving demands of big data. In response, this paper proposes an efficient local-global context fusion network (LGCF-Net) based on a Siamese architecture, aimed at enhancing the accuracy and efficiency of change detection in remote sensing images. LGCF-Net incorporates efficient local-global context aggregator (ELGCA) module and cross fusion attention module (CFAM) to effectively improve the performance of change detection. The proposed method is evaluated on the SYSU-CD dataset, achieving accuracy and recall of 93.21% and 95.39%, respectively, which confirms the effectiveness of the proposed method.