鲁棒的三正交灭点摄像机自标定算法  

Robust camera calibration algorithm with three-orthogonal vanishing point

作　　者：徐小泉 杨奇 李超 段伯韬 

机构地区：[1]63771部队

出　　处：《电子测量与仪器学报》2018年第4期65-73,共9页Journal of Electronic Measurement and Instrumentation

基　　金：国家高技术研究发展计划(863)资助项目

摘　　要：基于三正交灭点的摄像机标定算法对相对位姿敏感度高,采用随机获取的图像进行标定合格概率低,针对这个问题,提出了一种鲁棒的三正交灭点摄像机自标定算法。该算法通过求取内参数标定不确定度最小意义下的最优权值,对单幅图像三正交灭点标定结果进行加权,得到优化的标定结果;并在加权方法的基础上,引入随机采样策略剔除由于粗大误差或灭点分布退化引起的外点图像,提高多幅图像三正交灭点标定算法的稳定性和精度。仿真及实物实验结果证明算法标定的稳定性和精度较现有算法有显著提升。Camera calibration technique with three-orthogonal vanishing point is susceptible to relative pose,and has a low probability to give qualified calibration result with random taken images. Aiming at this problem,a robust camera calibration algorithm with threeorthogonal vanishing point is proposed. First the algorithm calculates the optimal calibration result by weighting the calibration results of independent images with three-orthogonal vanishing point,and the weights are obtained by minimizing calibration uncertanty of internal parameters. Besides that,RANSAC scheme is introduced to the algorithm to handle the outliers caused by degenerated distribution of vanishing points or cross noise,in order to further improve the stability and calibration accuracy. The experiment results with synthetic and real camera images indicate that the accuracy and robustness of the proposed algorithm is much better than existing algorithms.

关 键 词：摄像机自标定 三正交灭点 最小误差 随机采样 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]