RegionFuse：面向红外与可见光图像融合的区域自适应像素分布学习
RegionFuse: Region-Adaptive Pixel Distribution Learning for Infrared and Visible Image Fusion
将混合专家（MoE）与区域级注意力结合，按局部像素分布动态生成融合权重，这一「区域自适应权重生成」思路可直接迁移到多曝光融合、多模态特征融合、动态卷积等需要空间自适应加权的视觉任务。
原文依据 dynamically generates fusion weights based on local pixel distributions to construct a region-wise adaptive fusion paradigm
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论红外与可见光图像融合（IVIF）旨在结合红外与可见光图像中的互补信息，以克服单一模态的局限性。现有方法通常采用固定或样本自适应的融合范式，其融合权重是静态的或由全局像素分布导出，往往忽略了图像内部像素分布的空间不一致性，导致性能欠佳。为解决该问题，本文提出 RegionFuse，一种用于 IVIF 的区域自适应像素分布学习网络，它基于局部像素分布动态生成融合权重，从而构建区域级自适应融合范式。RegionFuse 引入区域混合注意力（MoRA）机制，将每个区域分配给若干专门的专家，从而实现针对特定局部分布的区域级特征交互。此外，本文设计了区域特征压缩模块（RFCM），并将其置于每个 MoRA 之后，以增强信息丰富的区域并抑制冗余区域。在多个基准上的大量实验证明了 RegionFuse 的优越性与鲁棒性，尤其是在处理非均匀像素分布时。在 NIR-VIS 及下游任务上的评估进一步证实了其泛化性与实用价值。代码已发布于 https://github.com/DarkIceField/RegionFuse。
英文原文摘要 1378 字符
Infrared and Visible Image Fusion (IVIF) aims to combine complementary information from infrared and visible images to overcome the limitations of a single modality. While existing methods typically employ fixed or sample-adaptive fusion paradigms where fusion weights are static or derived from global pixel distributions, they often overlook spatial inconsistencies in pixel distribution within images, leading to suboptimal performance. To address this issue, we propose RegionFuse, a Region-Adaptive Pixel Distribution Learning Network for IVIF, which dynamically generates fusion weights based on local pixel distributions to construct a region-wise adaptive fusion paradigm. RegionFuse introduces a Mixture of Region Attention (MoRA) mechanism, which assigns each region to several specialized experts, enabling region-level feature interaction tailored to specific local distributions. Furthermore, we design a Region Feature Compression Module (RFCM) and place it after each MoRA to enhance informative regions and suppress redundant ones. Extensive experiments on various benchmarks demonstrate the superiority and robustness of RegionFuse, especially in handling non-uniform pixel distributions. Evaluations on NIR-VIS and downstream tasks further confirm its generalizability and practical utility. The code is available at https://github.com/DarkIceField/RegionFuse.