lidar data processing

This work addresses the challenge of insufficient accuracy in extrinsic calibration between LiDAR and cameras in markerless scenarios, primarily caused by sparse cross-modal correspondences. To overcome this limitation, we propose a geometry-faithful joint optimization method based on 3D Gaussian splatting. By incorporating multi-view LiDAR depth supervision and freezing photometric gradients during differentiable rendering to prevent updates to Gaussian spatial parameters, our approach ensures that the proxy geometry remains consistent with the true LiDAR structure, thereby avoiding interference from rendering-based optimization on calibration accuracy. The method enables end-to-end extrinsic calibration and demonstrates consistently superior performance over existing markerless approaches on public autonomous driving datasets, achieving significant and stable improvements in calibration precision.

To address the challenges of colorization, chromatic distortion, and detail loss in mechanical LiDAR point clouds under low-light conditions, this paper proposes a tightly coupled multi-camera–LiDAR fusion framework. Methodologically, it constructs a 360° panoramic system comprising four synchronized cameras, integrating intrinsic calibration, automatic extrinsic estimation (target-free), inter-camera color consistency correction, and an embedded low-light image enhancement module; enhanced high-fidelity color images are then precisely projected onto Velodyne Puck Hi-Res point clouds. The key contribution is the first end-to-end integration of a deep low-light enhancement model into the sensor fusion pipeline, significantly improving color fidelity and texture discernibility in extremely dark scenes (<0.1 lux). Additionally, the framework achieves fully automatic geometric registration and color correction, enabling plug-and-play deployment. Experimental results demonstrate robust, high-fidelity, omnidirectional colored point cloud generation at near-real-time performance.

This work proposes the first adaptive LiDAR sensing framework that bridges the gap between data acquisition and downstream point cloud registration tasks. Unlike conventional LiDAR systems that employ fixed sensing parameters—often leading to redundant or insufficient data and high computational costs—our approach integrates registration performance feedback directly into the sensing process. By jointly optimizing acquisition parameters and registration hyperparameters in an end-to-end manner, the framework dynamically balances point cloud density, noise, and sparsity. Evaluated on the CARLA simulation platform, the method significantly outperforms fixed-parameter baselines, achieving higher registration accuracy and efficiency while maintaining strong generalization capabilities, thereby transcending the limitations of traditional static perception paradigms.

This paper addresses the challenge of accurately localizing the center of checkerboard targets in unstructured, sparse, and high-noise 3D point clouds—particularly those acquired by low-cost LiDAR sensors. To this end, we propose a robust measurement framework specifically designed for such data. Our method integrates 3D template matching, synthetic-data-driven target modeling, robust preprocessing of real-world point clouds, and sub-pixel-level center estimation into an end-to-end pipeline. A key innovation is the first application of color-enabled Iterative Closest Point (ICP) to template matching on unstructured point clouds, thereby overcoming the strong reliance of conventional approaches on point cloud ordering and low noise levels. Experimental results demonstrate sub-pixel localization accuracy on synthetic data and validate the method’s feasibility and robustness in practical applications—including point cloud registration, long-term structural monitoring, and multi-sensor fusion—using real LiDAR acquisitions.

LiDAR localization and mapping suffer from inaccurate pose estimation in geometrically degenerate environments—such as textureless regions or parallel surfaces—due to failure of point-to-plane optimization. To address this, we propose a probabilistic degeneration detection method based on noise propagation, which jointly models uncertainty in both point positions and surface normals, and integrates this into Hessian matrix uncertainty analysis. This enables real-time identification of degenerate directions and adaptive attenuation of pose updates. Our approach is interpretable and supports datasheet-driven quantification of degeneration probability. Evaluated on four real-world datasets, the method significantly improves registration robustness and accuracy over state-of-the-art approaches, particularly in severely degenerate scenarios.

This work addresses the challenges of poor sensor compatibility and low robustness to large initialization errors in LiDAR–camera extrinsic calibration under target-free scenarios. We propose a calibration method that requires neither calibration targets nor initial pose estimates. Given only a single RGB image and one LiDAR scan line (i.e., two 3D points), our approach leverages GlueStick to establish automatic 2D–3D point–line feature correspondences. An adaptive weighting scheme—based on geometric distribution—and an optimal feature selection strategy jointly guide a nonlinear optimization to estimate extrinsics. Evaluated across diverse LiDAR–camera configurations (e.g., Velodyne, Ouster, Livox with RGB cameras), our method achieves superior accuracy and robustness over state-of-the-art approaches, converging even with translation and rotation initialization errors up to ±1 m and ±30°. The implementation is open-sourced, significantly enhancing practical deployment efficiency.

This work addresses the challenge of reflection artifacts in terrestrial laser scanning (TLS) point clouds caused by glass surfaces in urban environments, which severely degrade downstream tasks. The authors propose a unified two-stage framework: first, they integrate multimodal vision foundation models with geometric cues to accurately estimate and complete glass regions; second, they introduce the RE-LGGS descriptor, grounded in the physical geometry of reflections, which leverages multiscale structural features and directional consistency to effectively identify and remove reflection artifacts. Moving beyond conventional symmetry assumptions, the method incorporates multimodal priors and a physics-driven local-global geometric similarity model. Evaluated on multiple public TLS datasets, the approach significantly outperforms existing methods, achieving substantial improvements in both accuracy and robustness of artifact removal.

This work addresses the system-level inconsistency in extrinsic calibration of multi-camera–LiDAR systems caused by independent estimation for each camera. To resolve this, the authors propose a two-stage joint calibration framework: first, the CMRNext network is employed to obtain initial extrinsics and 2D–3D correspondences for each camera–LiDAR pair; subsequently, a multi-frame bundle adjustment jointly optimizes all extrinsics by integrating reprojection errors with single-camera priors and inter-camera relative pose constraints, yielding globally consistent estimates. This approach uniquely combines learning-based pairwise initialization with explicit multi-camera geometric constraints. Evaluated on KITTI, it achieves a translation error of 0.89 cm and a rotation error of 0.038°, and on the Walkley dataset, it reduces translation error from 108.6 cm to 3.1 cm, significantly enhancing cross-domain robustness and calibration accuracy.

This work addresses the challenge of glare artifacts in solid-state LiDAR caused by internal multipath reflections, which manifest as “ghost” objects in point clouds and impair perception reliability. The authors propose, for the first time, a Transient Glare Spread Function (TGSF) model that characterizes internal glare as a scene-independent linear operator. Leveraging transient measurements from single-photon LiDAR and principles from linear system theory, the method enables real-time, training-free glare suppression directly at the waveform level, seamlessly integrating with existing signal processing pipelines. Through probabilistic inference, the approach effectively removes glare artifacts while fully preserving genuine scene geometry, significantly mitigating severe distortions observed in real hardware and thereby enhancing perception safety.

This work addresses the boundary blurring and distortion in LiDAR–camera extrinsic calibration caused by laser beam footprint effects and mixed-intensity returns. To this end, the authors propose a joint calibration method that integrates boundary response modeling. By co-observing visual fiducials on a printable planar calibration board and LiDAR-visible circular reflective boundaries, the approach iteratively refines 3D LiDAR edge feature points. It further incorporates an intensity- and geometry-constrained refinement strategy and a confidence-weighted reprojection optimization framework. Notably, this is the first method to embed explicit modeling of LiDAR boundary response characteristics into the extrinsic calibration pipeline. Evaluated on real-world data, it achieves sub-pixel reprojection accuracy and millimeter-level feature consistency, substantially improving downstream visual–LiDAR odometry performance.