{
  "paper_id": "91f4ec5b8e62d535beaa8e5f7a29afeb",
  "paper_title": "EasyColor: Reflectivity Assisted Dense Point Cloud RGB Colorizing Without Accurate Time Synchronization and Extrinsic Calibration",
  "paper_authors": [
    "Zhiyu Zhou",
    "Zhi Gao",
    "Yong Li",
    "Hang Zhen"
  ],
  "publication_date": "2025",
  "publication_venue": "IEEE Robotics and Automation Letters",
  "abstract": "A precise RGB-colored point cloud map serves as a reliable data foundation and a compact multi-sensor representation for tasks such as 3D perception, localization, and navigation. Therefore, efficient and accurate alignment between camera and LiDAR is essential. However, most existing color mapping methods require precise time synchronization and well-calibrated extrinsic value, and these colorizing frameworks are often hard to use. In this paper, we propose EasyColor, a precise colorization method that jointly leverages LiDAR reflectivity and visual constraint, effectively removing the need for time synchronization and extrinsic calibration between the two sensors. Our method follows a plug-and-play framework. Given an image sequence and a pre-built point cloud map, we first generate 2D reflectivity image sequence by exploiting variations in LiDAR reflectivity intensity. A cross-domain matching network, combined with a Perspective-n-Point (PnP) solver, is employed to align the reflectivity and camera image sequence. To further optimize camera poses and ensure photometric consistency, we fuse visual epipolar geometric constraint and PnP constraint. Our joint optimization produces both accurate and photometrically consistent colored map, along with optimized camera poses. Extensive experiments demonstrate that EasyColor achieves precise point cloud colorization without requiring synchronized LiDAR and camera data and extrinsic values. To facilitate future research, we will release EasyColor and dataset after review stage.",
  "keywords": [
    "Localization",
    "mapping",
    "sensor fusion"
  ],
  "paper_institutions": [
    "Wuhan University"
  ],
  "output_language": "zh",
  "provider": "doubao",
  "model": "glm-5.3",
  "prompt_template": "simple",
  "extracted_at": "2026-09-17T09:42:29.498882Z",
  "template_tag": "simple",
  "summary": "### 摘要\n\n精确的RGB彩色点云地图是三维感知、定位与导航任务的重要数据基础和紧凑的多传感器表示，但现有着色方法大多依赖相机与LiDAR间精确的时间同步和良好标定的外参：单目稠密SLAM着色精度不足，基于特征的方法（如Colmap-PCD、Koide的标定工具）在无纹理环境中不可靠且未利用帧间约束，基于运动的方法（如OmniColor）缺乏特征级对齐并受脆弱视觉里程计噪声影响。为此，本文提出即插即用的EasyColor框架：给定图像序列与预建点云地图，先将LiDAR点投影至图像平面生成2D反射率图像序列，并经直方图均衡化与gamma校正（γ=0.5）增强；随后采用跨域匹配网络MINIMA结合PnP求解器建立2D-3D对应，再融合视觉对极几何约束进行联合位姿优化，核心目标函数为 min E_pnp^i + E_pnp^(i+1) + λ·E_rot（λ=1，Levenberg-Marquardt求解），其中E_pnp为PnP重投影残差、E_rot为帧间相对旋转一致性误差；配合关键帧选择机制（内点比例阈值τ_s=0.6）、颜色渗色检测与不可靠边缘区域移除策略完成高质量着色。实验在公开数据集KITTI-360、FastLIVO2、MaRSLVIG及自采数据集（Velodyne VLP-16与全向相机，图像2Hz、LiDAR 10Hz）上开展，并人为加入0.1s时间偏移以验证无需严格时间同步，采用平移/旋转误差、重投影PSNR与SSIM作为指标：EasyColor将平移误差从Colmap-PCD的2%以上降至0.1–0.3%、旋转误差从10^-2降至约10^-3度/米，PSNR/SSIM在绝大多数序列上优于基线与Colmap-PCD（如hku1序列PSNR由15.86提升至20.54），消融实验证明了反射率增强、对极约束与关键帧机制的必要性；运行时分析显示每周期约1.98–2.47秒（跨域匹配最耗时），显示实时在线着色潜力。未来将探索与实时移动测量系统集成以实现在线彩色建图；实际应用中需具备含反射率信息的稠密点云地图并假设静态场景，当帧间快速运动、粗糙外参精度过低或时间偏移超过5秒时共视区域缩小会导致匹配失败，重复纹理等因素亦会引入误差。\n\n### 关键要点\n\n- EasyColor是即插即用框架，仅需预建点云地图（含反射率强度）与图像序列即可完成精确着色，无需相机-LiDAR时间同步与精确外参标定。\n- 方法将LiDAR点投影到图像平面生成2D反射率图像序列，并用直方图均衡化与gamma校正（γ=0.5）增强，显著改善原始反射率图像的空洞、噪声与低对比度问题。\n- 在比较SuperGlue、LoFTR、XFeat与MINIMA等匹配方法后选用MINIMA进行跨域图像匹配，其在精度与鲁棒性间取得最佳平衡，并借助索引图将2D-2D匹配转换为2D-3D对应关系。\n- 联合优化将LiDAR-相机PnP重投影残差与帧间对极几何相对旋转约束结合，目标为 min E_pnp^i + E_pnp^(i+1) + λ·E_rot（λ=1），以Levenberg-Marquardt算法求解，同时保证位姿精度与光度一致性。\n- 关键帧选择机制以复合内点比例 S_(i,i+1) 对候选帧对评分（阈值τ_s=0.6），仅将合格帧对送入联合优化与着色模块，抑制潜在外点影响。\n- 着色阶段提出颜色渗色检测（深度缓冲保留最小深度点）与不可靠边缘区域移除（3×3局部深度梯度与10×10邻域天空无效像素比例检测）两种策略，有效消除颜色渗色与边缘错误投影。\n- 实验在KITTI-360、FastLIVO2、MaRSLVIG公开数据集与自采数据集（Velodyne VLP-16+全向相机，图像2Hz、LiDAR 10Hz）上进行，并人为加入0.1s时间偏移以验证无需严格时间同步。\n- 位姿精度方面，EasyColor将平移误差从Colmap-PCD的2%以上降至0.1–0.3%，旋转误差从10^-2降至约10^-3度/米；消融实验显示移除反射率增强使误差升高30–50%，移除对极约束使旋转精度下降最高达一个数量级。\n- 重投影PSNR/SSIM评估显示EasyColor在大多数序列上优于基线与Colmap-PCD，例如FastLIVO2 hku1序列PSNR从15.86提升至20.54、SSIM从0.66提升至0.84。\n- 运行时分析（Intel i9-12900KF、NVIDIA RTX 4060）显示每个处理周期总耗时约1.98秒与2.47秒，跨域匹配最耗时（1.18/1.29秒），其余模块均在0.5秒内，具备实时在线着色潜力。\n- 潜在失败情形包括帧间快速运动、粗糙外参精度过低或时间偏移大于5秒导致共视区域缩小与匹配错误，以及重复纹理等损害图像匹配的因素。\n- 未来工作计划将EasyColor与实时移动测量系统集成以实现在线彩色建图，作者承诺在评审阶段后发布代码与数据集。"
}