Targetless Extrinsic Calibration for LiDAR and Camera Systems with Intensity Information

Accurate and efficient external calibration, available in real-time within a targetless environment, remains a significant challenge due to the inherent inhomogeneous reflectance, density distribution, and small FOV of Non-Repeat Scanning (NRS) LiDAR system data, these issues bring difficulties for subsequent applications in areas such as simultaneously localization and mapping (SLAM) and autonomous driving. To address these issues, this paper proposes a targetless method for the automatic calibration of LiDAR and camera systems. Initially, we analyzed the complexity and variability of LiDAR point cloud information and developed a feature extraction method that integrates geometric and intensity data. Subsequently, we optimized the intensity data within the point cloud to extract reliable intensity information. Finally, we matched and iteratively optimized the sensor system data to obtain the final extrinsic parameters. Through extensive experimental analysis, we verified the effectiveness and accuracy of our algorithm.

I. Introduction

In mobile robotics and autonomous driving, precise positioning and navigation are crucial. Cameras provide rich color information, while LiDAR offers direct 3D measurements. Combining these sensors through multi-sensor fusion enhances performance, leveraging their complementary strengths 0. LiDAR and optical cameras are widely used together in SLAM [1] , autonomous driving [2] , and target detection [3] . However, inaccurate external parameters can lead to disastrous outcomes in the application of subsequent tasks, which increases the demand for precise external calibration.