Improved Remote Sensing Change Detection Through Adaptive Boundary Enhancement

By comparing images from different time periods, remote sensing change detection (RSCD) effectively monitors dynamic changes on Earth's surface and has been applied in various fields such as natural assessment, urban expansion analysis, and military operations. However, multi-temporal change detection faces challenges such as pseudo-changes caused by seasonal vegetation variations and color discrepancies. While deep learning methods have demonstrated excellent performance in enhancing detection accuracy and robustness, existing methods still face limitations in edge enhancement, making it difficult to generate high-quality remote sensing results. Therefore, we propose a lightweight RSCD method that incorporates an adaptive boundary enhancement module. This module significantly improves the generation of edge features through weighted concatenation, thereby increasing the detection rate of real changes. Experimental results show that the proposed method exhibits stronger adaptability and superior generalization capability in complex scenarios.