GMBFormer: 一种NDVI引导的全局记忆库Transformer用于超高分辨率影像城市绿地提取
GMBFormer: An NDVI-Guided Global Memory Bank Transformer for Urban Green-Space Extraction from Ultra-High-Resolution Imagery
AI总结：
针对超高分辨率影像城市绿地提取中视觉相似植被模式语义复用受限及NDVI与RGB特征融合模糊的问题，提出GMBFormer框架，通过解耦NDVI作为物理门控并利用全局记忆库进行选择性原型检索，在三个数据集上提升了分割精度。
摘要：
从超高分辨率（UHR）影像中提取城市绿地通常逐块进行，这限制了空间分离但视觉相似的植被模式之间的语义复用。将归一化差异植被指数（NDVI）直接注入红绿蓝（RGB）主干网络也会模糊视觉外观学习与物理植被置信度的作用。我们提出了GMBFormer，一个基于SegFormer的框架，用选择性、相似性驱动的原型检索替代邻域驱动的特征传播。只有RGB通道进入主干网络和解码器，而NDVI被解耦为一个物理信息门控，通过动量更新将高置信度植被描述符纳入紧凑的全局记忆库。在训练和推理过程中，当前块通过记忆介导的交叉注意力查询存储的原型，并以有限的开销集成检索到的响应。实验使用了自建的成都UHR数据集（含7,700个标注的512×512块）以及从公共国际摄影测量与遥感学会（ISPRS）波茨坦数据集派生的两种减少标签设置。在相同的训练和评估协议下，GMBFormer分别获得了89.25%/94.31%、92.17%/95.92%和83.72%/90.86%的平均交并比（mIoU）/平均Dice（mDice）分数，在每种设置下均优于受控的SegFormer-B4基线。消融研究表明，解耦的NDVI准入、记忆检索、容量和动量共同决定了最终性能。
英文摘要：
Urban green-space extraction from ultra-high-resolution (UHR) imagery is commonly performed patch by patch, which limits semantic reuse among spatially separated but visually similar vegetation patterns. Directly injecting the Normalized Difference Vegetation Index (NDVI) into red-green-blue (RGB) backbones can also blur the roles of visual appearance learning and physical vegetation confidence. We propose GMBFormer, a SegFormer-based framework that replaces adjacency-driven feature propagation with selective, similarity-driven prototype retrieval. Only RGB channels enter the backbone and decoder, while NDVI is decoupled as a physics-informed gate that admits high-confidence vegetation descriptors into a compact global memory bank through momentum updates. During training and inference, the current patch queries stored prototypes through memory-mediated cross-attention, and the retrieved response is integrated with bounded overhead. Experiments use a self-constructed Chengdu UHR dataset with 7,700 labeled 512 x 512 patches and two reduced-label settings derived from the public International Society for Photogrammetry and Remote Sensing (ISPRS) Potsdam dataset. Under the same training and evaluation protocol, GMBFormer obtains mean intersection over union (mIoU)/mean Dice (mDice) scores of 89.25%/94.31%, 92.17%/95.92%, and 83.72%/90.86%, respectively, improving the controlled SegFormer-B4 baseline in each setting. Ablation studies indicate that decoupled NDVI admission, memory retrieval, capacity, and momentum jointly shape the final performance.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
