基于棋盘标定板的优化相机参数标定方法
Optimized Camera Parameter Calibration Method Based on Checkerboard
林绿开 1钮倩倩 1李毅2
作者信息
- 1. 温州大学 计算机与人工智能学院,浙江 温州 325000
- 2. 温州大学 计算机与人工智能学院,浙江 温州 325000;浙江大学 计算机科学与技术学院,浙江 杭州 310058
- 折叠
摘要
相机标定是计算机视觉与重建定位的重要基础,是构建二维图像与三维空间联系的重要桥梁.该文根据传统线性相机成像的规律,在理想的针孔相机成像模型的基础上,同时针对张正友标定法中相机参数求解过程较为复杂与繁琐的问题,采用一种简易的标定方法来求解单目相机的相机参数.该方法在平面张正友标定方法的基础上,通过预先估计部分相机参数的初值进而求解理想针孔相机成像模型后,再进一步优化预估参数的方式,根据棋盘标定板的特征点的空间坐标与图像坐标的对应方程,利用奇异值分解和Levenberg-Marquardt优化的方法进行参数求解,从而避免求解繁杂的约束矩阵后再分别对相机参数进行求解的过程,达到直接求解相机参数的目的,优化求解过程,最终实现单目相机的相机标定.实验结果表明,使用该方法标定的单目相机在物距一米以下时的标定误差能够满足绝大部分视觉与定位的应用需求,且该方法具有较好的稳定性和可行性.
Abstract
Camera calibration is an essential basis for computer vision and reconstruction positioning,which is a critical bridge between two-dimensional images and three-dimensional space.According to the regulation of traditional linear camera imaging and the ideal pinhole camera imaging model,a simple calibration method is proposed to solve the camera parameters of the monocular camera,which can address the complex and tedious calculation problem based on the plane Zhengyou Zhang Calibration Method.The method is based on the plane Zhengyou Zhang Calibration Method,which optimizes the predicted parameters by pre-estimating the initial values of some camera parameters and solving the ideal pinhole camera imaging model.According to the equations of the spatial and image coordinates of the checkerboard calibration board's feature points,the singular value decomposition and Levenberg-Marquardt optimization methods are used for parameter solution,which avoids the process of calculating the complicated constraint matrix before the camera parameters separately,thus achieving the purpose of solving the camera parameters directly,while optimizing the solving process and finally realizing camera calibration of monocular cameras.The experimental results show that the calibration error of the monocular camera calibrated with the proposed method can satisfy most vision and positioning applications at object distances below one meter,with high stability and feasi-bility.
关键词
单目相机/标定算法/奇异值分解/Levenberg-Marquardt优化/张正友标定法Key words
monocular camera/calibration algorithm/singular value decomposition/Levenberg-Marquardt optimization/Zhengyou Zhang Calibration Method引用本文复制引用
基金项目
浙江省软科学研究计划重点项目(2022C25033)
温州市科技计划项目(R20200025)
出版年
2023
国家科技期刊平台
NSTL
维普
万方数据