Multi-Scale Feature Fusion Module for Change Detection in Remote Sensing Images

Change detection refers to an identification of pixels that have undergone semantic changes among two remote sensing images captured at different times in a same area. The problem statement is that remote sensing images has high-resolution causes a challenge for current deep learning-based change and missed detection of small targets. It is overcome by the proposed method with Multi-scale Feature Fusion (MSFF) for feature fusion; semantic variation data is extracted from the Global Semantic Aggregation Module (GSAM); during a decoding phase, Attention Fusion Module (AFM) is applied to extract features; and Cross-scale Fusion Module (CFM) is used to enhance accuracy by removing background noise and the Levir-CD dataset is used. The numerical validation of this approach has a precision of 95.43%, a recall of 97.80%, an IoU of 93.49%, an OA of 99.67%, and a Kappa of 96.59%. The Attention-Guided Siamese Network (SAGNet), Multi-Stage Progressive Change Detection Network (MSP-CD), Teacher-Student Framework (TSF), and Global Attention Module (GAM) are the existing models; the limitations of these methods are overcome by the proposed method.