基于球面角点特征的鱼眼相机与激光雷达联合标定方法  

Joint calibration method of fisheye camera and LiDAR based on spherical corner features

作　　者：高志远[1,2] 邢玉波 肖尧 史航 解杨敏 GAO Zhiyuan;XING Yubo;XIAO Yao;SHI Hang;XIE Yangmin(School of Mechanical and Electrical Engineering and Automation,Shanghai University,Shanghai 200444,China;Shanghai Key Laboratory of Intelligent Manufacturing and Robotics,Shanghai University,Shanghai 200444,China)

机构地区：[1]上海大学机电工程与自动化学院,上海200444 [2]上海大学上海市智能制造及机器人重点实验室,上海200444

出　　处：《现代电子技术》2024年第13期102-107,共6页Modern Electronics Technique

基　　金：上海市自然科学基金面上项目:基于多模融合感知的复杂三维环境语义辅助稠密SLAM(20ZR1419100)。

摘　　要：针对鱼眼相机和激光雷达的联合标定问题,提出一种基于球面角点特征的鱼眼相机与激光雷达联合标定方法。首先利用静电斥力物理模型得到了均匀分布在球面上的点集,并以这些点集为生长点对球面进行Voronoi划分,得到了均匀的球面像素,实现了鱼眼相机图像在球面上的均匀重采样;之后在球面上提取了鱼眼相机图像的角点特征,有效地避免了图像畸变带来的角点提取困难的问题;最后利用17组标定数据在球面上实现了鱼眼相机和激光雷达联合标定,在鱼眼图像上的重投影误差小于2个像素,并通过数据融合实验复现了真实场景中的彩色三维效果。A joint calibration method of fisheye camera and LiDAR based on spherical corner features is proposed to address the issue of joint calibration of fisheye camera and LiDAR.The point sets uniformly distributed on the sphere are obtained by the electrostatic repulsion physical model.The sphere is subjected to Voronoi partitioning with these point sets as growing points,and then the uniform spherical pixels are obtained,which realizes the uniform resampling of the fisheye camera image on the sphere.And then,the corner features of the fisheye camera image are extracted on the sphere,which effectively avoids the difficult corner extraction caused by image distortion.17 sets of calibration data are used to achieve joint calibration of the fisheye camera and LiDAR on the sphere.The error of reprojection on the fisheye image is less than 2 pixels,and the color 3D effect in the real scene is reproduced by data fusion experiments.

关 键 词：鱼眼相机 激光雷达 像素采样 VORONOI划分 球面角点 联合标定 三维重建 

分 类 号：TN29-34[电子电信—物理电子学]