一种面向增强现实中相机标定的天牛须搜索改进算法  

An improved beetle antennae search algorithm for camera calibration in augmented reality

作　　者：张秀山[1] 荣晗戈 徐浩 黄健 李永杰[1] ZHANG Xiushan;RONG Hange;XU Hao;HUANG Jian;LI Yongjie(College of Electronics Engineering,Naval Univ.of Engineering,Wuhan 430033,China;Unit No.32383,Wuhan 430300,China;Unit No.91550,Dalian 116000,China)

机构地区：[1]海军工程大学电子工程学院,武汉430033 [2]32383部队,武汉430300 [3]91550部队,辽宁大连116000

出　　处：《海军工程大学学报》2023年第5期40-45,共6页Journal of Naval University of Engineering

基　　金：国家部委基金资助项目(HCX-2020-63)。

摘　　要：针对三维注册初始化中内参难以进行标定的问题,分析了亚像素精度与相机成像模型,并对天牛须搜索算法进行了改进,在此基础上提出了一种求解三维注册初始内参的相机标定方法。首先,在图片中提取角点并进行亚像素精化;然后,在相机成像方程中引入畸变系数,重新推导成像方程并求解内参;最后,提出天牛须扰动搜索算法并对内参进行非线性优化。实验结果表明,该方法可明显减小相机的重投影误差,可以为三维注册模块提供初始化内参。Aiming at the problem that it is difficult to calibrate internal parameters in the initialization of 3D registration,the sub-pixel accuracy and camera imaging model was analyzed,the beetle antennae search algorithm was improved.A camera calibration method for solving initial internal parameters of 3D registration was proposed.Firstly,the sub-pixel corner points in the original image were extracted and refined.Furthermore,distortion coefficient was introduced into the camera imaging equation and internal parameters were obtained by new imaging equation.Finally,beetle antennae perturbation search algorithm was proposed for nonlinear optimization of internal parameters.Experimental results show that the reprojection error is reduced,the method can provide an internal parameters for initialization of the 3D registration module.

关 键 词：增强现实 相机标定 亚像素 成像模型 天牛须搜索算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]