LiDAR as a Geometric Prior: Enhancing Camera Pose Tracking Through High-Fidelity View Synthesis

This paper presents a robust framework for monocular camera pose estimation by leveraging high-fidelity, pre-built 3D LiDAR maps. The core of our approach is a render-and-match pipeline that synthesizes photorealistic views from a dense LiDAR point cloud. By detecting and matching keypoints between...

Full description

Saved in:
Bibliographic Details
Main Authors: Rafael Muñoz-Salinas, Jianheng Liu, Francisco J. Romero-Ramirez, Manuel J. Marín-Jiménez, Fu Zhang
Format: Article
Language:English
Published: MDPI AG 2025-08-01
Series:Applied Sciences
Subjects:
Online Access:https://www.mdpi.com/2076-3417/15/15/8743
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849766017221263360
author Rafael Muñoz-Salinas
Jianheng Liu
Francisco J. Romero-Ramirez
Manuel J. Marín-Jiménez
Fu Zhang
author_facet Rafael Muñoz-Salinas
Jianheng Liu
Francisco J. Romero-Ramirez
Manuel J. Marín-Jiménez
Fu Zhang
author_sort Rafael Muñoz-Salinas
collection DOAJ
description This paper presents a robust framework for monocular camera pose estimation by leveraging high-fidelity, pre-built 3D LiDAR maps. The core of our approach is a render-and-match pipeline that synthesizes photorealistic views from a dense LiDAR point cloud. By detecting and matching keypoints between these synthetic images and the live camera feed, we establish reliable 3D–2D correspondences for accurate pose estimation. We evaluate two distinct strategies: an Online Rendering and Tracking method that renders views on the fly, and an Offline Keypoint-Map Tracking method that precomputes a keypoint map for known trajectories, optimizing for computational efficiency. Comprehensive experiments demonstrate that our framework significantly outperforms several state-of-the-art visual SLAM systems in both accuracy and tracking consistency. By anchoring localization to the stable geometric information from the LiDAR map, our method overcomes the reliance on photometric consistency that often causes failures in purely image-based systems, proving particularly effective in challenging real-world environments.
format Article
id doaj-art-5a15987c741a42c8a4745051e4c417a0
institution DOAJ
issn 2076-3417
language English
publishDate 2025-08-01
publisher MDPI AG
record_format Article
series Applied Sciences
spelling doaj-art-5a15987c741a42c8a4745051e4c417a02025-08-20T03:04:42ZengMDPI AGApplied Sciences2076-34172025-08-011515874310.3390/app15158743LiDAR as a Geometric Prior: Enhancing Camera Pose Tracking Through High-Fidelity View SynthesisRafael Muñoz-Salinas0Jianheng Liu1Francisco J. Romero-Ramirez2Manuel J. Marín-Jiménez3Fu Zhang4Departamento de Informática e Inteligencia Artificial, Edificio Albert Einstein, Campus de Rabanales, Universidad de Córdoba, 14071 Córdoba, SpainMechatronics and Robotic Systems (MaRS) Laboratory, Department of Mechanical Engineering, University of Hong Kong (HKU), Hong Kong, ChinaDepartamento de Teoría de la Señal y Comunicaciones y Sistemas Telemáticos y Computación, Campus de Fuenlabrada, Universidad Rey Juan Carlos, 28942 Fuenlabrada, SpainDepartamento de Informática e Inteligencia Artificial, Edificio Albert Einstein, Campus de Rabanales, Universidad de Córdoba, 14071 Córdoba, SpainMechatronics and Robotic Systems (MaRS) Laboratory, Department of Mechanical Engineering, University of Hong Kong (HKU), Hong Kong, ChinaThis paper presents a robust framework for monocular camera pose estimation by leveraging high-fidelity, pre-built 3D LiDAR maps. The core of our approach is a render-and-match pipeline that synthesizes photorealistic views from a dense LiDAR point cloud. By detecting and matching keypoints between these synthetic images and the live camera feed, we establish reliable 3D–2D correspondences for accurate pose estimation. We evaluate two distinct strategies: an Online Rendering and Tracking method that renders views on the fly, and an Offline Keypoint-Map Tracking method that precomputes a keypoint map for known trajectories, optimizing for computational efficiency. Comprehensive experiments demonstrate that our framework significantly outperforms several state-of-the-art visual SLAM systems in both accuracy and tracking consistency. By anchoring localization to the stable geometric information from the LiDAR map, our method overcomes the reliance on photometric consistency that often causes failures in purely image-based systems, proving particularly effective in challenging real-world environments.https://www.mdpi.com/2076-3417/15/15/8743camera pose estimationLiDAR mapping3D reconstructionvisual SLAM
spellingShingle Rafael Muñoz-Salinas
Jianheng Liu
Francisco J. Romero-Ramirez
Manuel J. Marín-Jiménez
Fu Zhang
LiDAR as a Geometric Prior: Enhancing Camera Pose Tracking Through High-Fidelity View Synthesis
Applied Sciences
camera pose estimation
LiDAR mapping
3D reconstruction
visual SLAM
title LiDAR as a Geometric Prior: Enhancing Camera Pose Tracking Through High-Fidelity View Synthesis
title_full LiDAR as a Geometric Prior: Enhancing Camera Pose Tracking Through High-Fidelity View Synthesis
title_fullStr LiDAR as a Geometric Prior: Enhancing Camera Pose Tracking Through High-Fidelity View Synthesis
title_full_unstemmed LiDAR as a Geometric Prior: Enhancing Camera Pose Tracking Through High-Fidelity View Synthesis
title_short LiDAR as a Geometric Prior: Enhancing Camera Pose Tracking Through High-Fidelity View Synthesis
title_sort lidar as a geometric prior enhancing camera pose tracking through high fidelity view synthesis
topic camera pose estimation
LiDAR mapping
3D reconstruction
visual SLAM
url https://www.mdpi.com/2076-3417/15/15/8743
work_keys_str_mv AT rafaelmunozsalinas lidarasageometricpriorenhancingcameraposetrackingthroughhighfidelityviewsynthesis
AT jianhengliu lidarasageometricpriorenhancingcameraposetrackingthroughhighfidelityviewsynthesis
AT franciscojromeroramirez lidarasageometricpriorenhancingcameraposetrackingthroughhighfidelityviewsynthesis
AT manueljmarinjimenez lidarasageometricpriorenhancingcameraposetrackingthroughhighfidelityviewsynthesis
AT fuzhang lidarasageometricpriorenhancingcameraposetrackingthroughhighfidelityviewsynthesis