MIND-Calib: Multi-View, Intensity and Depth-Driven Dense 2D�3D Alignment for Single-Frame LiDAR�Camera Extrinsic Calibration
Shezhong Liu, Zibin Chen
AI summary
Problem
Existing targetless LiDAR-camera calibration methods struggle with time-consuming data preparation, poor robustness under sparse single-frame inputs, and limited generalization across different LiDAR architectures.
Approach
The method projects a single LiDAR scan into multiple virtual viewpoints to generate complementary depth and intensity images, uses cross-modal image matching to establish correspondences, and applies geometry-based densification to create a dense point cloud for precise 2D–3D alignment.
Key results
- Average 2.85 cm translation and 0.20° rotation accuracy across diverse LiDAR types
- True single-frame, zero-preparation calibration without trajectories or initialization
- Stable accuracy and robustness under sparse point cloud inputs
- Strong generalization across mechanical, solid-state, and MEMS LiDAR architectures
Why it matters
Provides a reliable, plug-and-play calibration solution for autonomous driving and robotics systems that need rapid, accurate sensor alignment across varying hardware and environments.
Abstract
Extrinsic calibration between LiDAR and camera is a crucial step in multi-sensor fusion, where targetless ap- proaches have attracted increasing attention for their flexibility and reusability. However, existing methods still suffer from three major limitations: time-consuming data preparation, lack of robustness under sparse single-frame input, and limited generalization across diverse LiDAR architectures. We pro- pose MIND-Calib, a truly single-frame, targetless calibration framework. The method generates depth and intensity images through virtual multi-view projection, and performs image- domain completion and back-projection to densify the point cloud and construct sub-pixel 2D–3D correspondences. High- precision extrinsics are then estimated via dual-channel cross- modal matching that leverages both depth and intensity modali- ties. Experiments on three representative LiDAR types (MEMS- based, solid-state, and mechanical spinning) as well as on public datasets demonstrate an average accuracy of 2.85 cm (with respect to an average scene depth of 40 meters) in translation and 0.20◦in rotation. More importantly, MIND-Calib not only achieves true single-frame calibration without any additional preparation, but also maintains stable accuracy under sparse inputs and exhibits strong generalization and robustness across devices and challenging environments.