ULF-Loc：面向 3D 高斯泼溅鲁棒视觉定位的无偏地标特征
ULF-Loc: Unbiased Landmark Feature for Robust Visual Localization with 3D Gaussian Splatting
论文揭示了 3DGS 中 alpha-blending 优化会给高斯点特征引入偏差，并提出几何加权特征融合替代有偏的特征优化；对做 3DGS 特征场、2D-3D 匹配定位的工作，这是一个可直接复用的特征学习范式修正。
原文依据 the widely adopted alpha-blending optimization inherently introduces bias into 3D point features
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论视觉定位是增强现实与自主导航的核心技术。近期方法将 3D 高斯泼溅（3DGS）的高效渲染与基于特征的定位相结合。这些方法依赖 2D 查询特征与 3D 高斯特征场之间的直接匹配，但由于所学高斯特征存在固有偏差，这往往导致误匹配。我们从理论上分析了 3DGS 中的特征学习过程，揭示出被广泛采用的 alpha 混合优化会固有地向 3D 点特征引入偏差。该偏差源于单个高斯与其邻近高斯之间的纠缠，使得所学特征不适合精确匹配任务。受此发现启发，我们提出 ULF-Loc，一个无偏地标特征框架，以几何加权特征融合替代有偏的特征优化。我们进一步引入关键点共识地标采样以选取可靠的高斯，并采用局部几何一致性验证来剔除由渲染伪影导致的误匹配。在 Cambridge Landmarks 数据集上，ULF-Loc 相比当前最优方法将平均中位平移误差降低 17%，同时效率更优，训练时间仅为 STDLoc 的 1/10，GPU 显存仅为 1/6。我们的代码见 https://github.com/Cyril-gyd/ULF-Loc。
英文原文摘要 1384 字符
Visual localization is a core technology for augmented reality and autonomous navigation. Recent methods combine the efficient rendering of 3D Gaussian Splatting (3DGS) with feature-based localization. These methods rely on direct matching between 2D query features and the 3D Gaussian feature field, but this often results in mismatches due to an inherent bias in the learned Gaussian feature. We theoretically analyze the feature learning process in 3DGS, revealing that the widely adopted alpha-blending optimization inherently introduces bias into 3D point features. This bias stems from the entanglement between individual Gaussians and their neighboring Gaussians, making the learned features unsuitable for precise matching tasks. Motivated by these findings, we propose ULF-Loc, an unbiased landmark feature framework that replaces biased feature optimization with geometry-weighted feature fusion. We further introduce keypoint-consensus landmark sampling to select reliable Gaussians and local geometric consistency verification to reject mismatches caused by rendering artifacts. On the Cambridge Landmarks dataset, ULF-Loc reduces the mean median translation error by 17% compared to the state-of-the-art, while achieving superior efficiency with only 1/10 the training time and 1/6 the GPU memory of STDLoc. Our code is accessible at https://github.com/Cyril-gyd/ULF-Loc.