标题： 精确的时空特征融合用于鲁棒的遥感变化检测 显示英文标题

标题： Precision Spatio-Temporal Feature Fusion for Robust Remote Sensing Change Detection

摘要： 遥感变化检测对于监测环境和城市变迁至关重要，但面临手动特征提取和对噪声敏感等挑战。 传统方法和早期深度学习模型，如卷积神经网络（CNN），难以捕捉复杂场景中准确变化检测所需的距离依赖关系和全局上下文。 虽然基于Transformer的模型缓解了这些问题，但其计算复杂度限制了其在高分辨率遥感中的应用。 在ChangeMamba架构的基础上，该论文提出了精确融合块，以捕捉通道级时间变化和像素级差异，实现细粒度的变化检测。 一种增强的解码器流程，结合轻量级通道缩减机制，在最小计算成本下保留局部细节。 此外，一种结合交叉熵、Dice和Lovasz目标的优化损失函数解决了类别不平衡问题，并提升了交并比（IoU）。 在SYSU-CD、LEVIR-CD+和WHU-CD数据集上的评估表明，与最先进的方法相比，该方法在精度、召回率、F1分数、IoU和总体准确性方面表现更优，突显了该方法在遥感变化检测中的鲁棒性。 为了完全透明，代码和预训练模型可在 https://github.com/Buddhi19/MambaCD.git 获取。 显示英文摘要

摘要： Remote sensing change detection is vital for monitoring environmental and urban transformations but faces challenges like manual feature extraction and sensitivity to noise. Traditional methods and early deep learning models, such as convolutional neural networks (CNNs), struggle to capture long-range dependencies and global context essential for accurate change detection in complex scenes. While Transformer-based models mitigate these issues, their computational complexity limits their applicability in high-resolution remote sensing. Building upon ChangeMamba architecture, which leverages state space models for efficient global context modeling, this paper proposes precision fusion blocks to capture channel-wise temporal variations and per-pixel differences for fine-grained change detection. An enhanced decoder pipeline, incorporating lightweight channel reduction mechanisms, preserves local details with minimal computational cost. Additionally, an optimized loss function combining Cross Entropy, Dice and Lovasz objectives addresses class imbalance and boosts Intersection-over-Union (IoU). Evaluations on SYSU-CD, LEVIR-CD+, and WHU-CD datasets demonstrate superior precision, recall, F1 score, IoU, and overall accuracy compared to state-of-the-art methods, highlighting the approach's robustness for remote sensing change detection. For complete transparency, the codes and pretrained models are accessible at https://github.com/Buddhi19/MambaCD.git