Evaluation of Front-End Feature Matching for Lunar Visual SLAM

📄 arXiv: 2610.30774v1 📥 PDF

作者: Guillem Casadesus Vila, Grace Gao

分类: cs.CV

发布日期: 2026-10-22

备注: 16 pages, 8 figures, 5 tables. Submitted to the 2027 IEEE Aerospace Conference


💡 一句话要点

评估前端特征匹配以解决月球视觉SLAM问题

🎯 匹配领域: 支柱三:空间感知与语义 (Perception & Semantics) 支柱六:视频提取与匹配 (Video Extraction)

关键词: 视觉SLAM 特征匹配 月球探测 视觉里程计 深空探索 机器学习 图像处理

📋 核心要点

  1. 现有的特征匹配方法主要针对城市和室内场景,缺乏对月球低纹理环境的适应性,导致视觉SLAM性能不足。
  2. 本文提出了对11种特征检测-匹配组合的基准测试,特别关注在月球测试床和类地数据集上的表现。
  3. 实验结果显示,SuperPoint在多个测试场景中表现优于SIFT,尤其在低光照条件下,具有更高的准确性和稳定性。

📝 摘要(中文)

月球探测车在导航卫星服务部署之前,主要依赖相机进行定位,并在之后的相对导航、危险检测和表面操作中继续使用相机。视觉里程计(VO)和同时定位与地图构建(SLAM)主要通过匹配帧间特征来估计运动。然而,现有的特征匹配器主要是在城市、室内和驾驶场景下开发的,缺乏对低纹理、光照条件恶劣的月球表面图像的比较。本文基准测试了11种经典和学习型特征检测-匹配组合,评估其在不同数据集上的表现,结果表明SuperPoint在月球测试床上优于SIFT,且在所有八个顺序数据集中保持跟踪能力。

🔬 方法详解

问题定义:本文旨在解决现有特征匹配方法在月球低纹理和光照条件下的不足,特别是缺乏对真实月球图像的有效匹配评估。

核心思路:通过基准测试11种经典和学习型特征检测-匹配组合,评估其在不同环境下的表现,以找到最适合月球探测的特征匹配方法。

技术框架:研究采用了多种数据集,包括合成数据、月球测试床和类地数据,评估了特征匹配的准确性和视觉里程计的轨迹误差。

关键创新:最重要的创新在于对SuperPoint与SIFT等经典方法的比较,发现SuperPoint在特定条件下表现优越,尤其是在低光照环境中。

关键设计:实验中使用了多种参数设置和匹配策略,SuperPoint与LightGlue的组合在速度上表现最佳,且在不同光照条件下的表现稳定。实验还表明,合成数据的结果不能完全代表真实环境的性能。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

实验结果显示,SuperPoint在月球测试床上优于SIFT,且在所有八个顺序数据集中保持跟踪能力。SuperPoint与LightGlue组合的速度为每帧16毫秒,显示出在特征匹配中的高效性和准确性。

🎯 应用场景

该研究的潜在应用领域包括月球探测、行星探索和其他低纹理环境下的视觉导航。通过优化特征匹配算法,可以提升探测车在复杂环境中的自主导航能力,推动深空探索的技术进步。

📄 摘要(原文)

Lunar rovers will primarily rely on cameras for localization until navigation satellite services are deployed, and will continue to need them afterwards for relative navigation, hazard detection, and surface operations. Visual odometry (VO) and Simultaneous Localization and Mapping (SLAM) mostly estimate motion by matching features between frames. However, feature matchers are developed on urban, indoor and driving scenes, planetary studies compare them on orbital registration or within complete SLAM systems, and real lunar imagery with reference poses is scarce, so matching between rover frames on low-texture, harshly lit terrain, and its effect on VO, has not been compared. We benchmark eleven classical and learned detector-matcher combinations on nine synthetic, lunar-testbed, and terrestrial-analogue datasets, and a feature tracker, scoring pairwise pose accuracy against baseline and the trajectory error of the resulting VO. Under a common rotation-only metric, the SuperPoint pairings outperform SIFT with ratio-tested FLANN matching on both lunar testbeds and one analogue site, with smaller or unresolved differences on synthetic data. When rotation and translation direction are scored jointly, this SIFT pairing exceeds both SuperPoint pairings on four synthetic datasets, so synthetic results alone do not establish relative performance on real rover imagery. In stereo VO, the SuperPoint pairings are not consistently more accurate than SIFT, but only they keep tracking on all eight sequential datasets. Pairings with classical detectors depend on lighting while SuperPoint changes little: a low spotlight makes them more accurate than diffuse light on the same testbed views, consistent with more contrast and detected keypoints. SuperPoint with LightGlue is the fastest learned combination (16 ms per frame), and DINOv2 retrieval proposals still need geometric verification.