散焦模糊量估计的相机加权标定方法  

Weighted Camera Calibration Method Considering Defocus Amount

作　　者：汪蕾 刘涛[1] 董琦聪 冯结青[1] Wang Lei;Liu Tao;Dong Qicong;Feng Jieqing(State Key Laboratory of CAD&CG,Zhejiang University,Hangzhou 310058)

机构地区：[1]浙江大学CAD&CG国家重点实验室,杭州310058

出　　处：《计算机辅助设计与图形学学报》2020年第3期410-417,共8页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(61732015,61472349);浙江省重点研发计划项目(2018C01090).

摘　　要：相机标定在计算机视觉领域中有着至关重要的作用.绝大多数相机标定方法假设相机为针孔模型,且需要良好聚焦的图像来保证相机内外参估计的准确性.然而,这些条件会受到相机景深的影响.在薄透镜相机模型假设下,提出了一种加权相机标定的方法,其权重考虑了控制点的模糊量信息.首先对棋盘格标定物上的每一个角点进行散焦模糊量估计,在标定过程中,将散焦模糊量的大小作为一个权重加入到标定能量函数最小化过程中,使得标定精度得到提高.该方法简单高效,不需要额外的数码设备或者特别定做的标定物.在Intel Core i7处理器的计算机下,使用合成数据以及真实数据上进行的实验结果表明,文中方法能够有效减小重投影误差,提高张正友标定方法的标定精度.Camera calibration plays a critical role in the field of computer vision. Most of the camera calibration methods assume the pinhole camera model and require well-focused images to ensure precise estimation of the intrinsic and extrinsic camera parameters. However, they are often impaired by lens systems with limited depth of field(DoF). In this paper, we propose a weighted camera calibration method that takes the defocus amount of control points into account. Under the assumption of a thin-lens camera model, blur estimation is performed for each individual control point on the calibration target with a checkerboard pattern. The defocus amount of individual control points participates in the calibration procedure in a weighted form to improve the accuracy of the results. No additional device or special calibration target is needed. Experiments were conducted using both synthetic and real images, and the results showed that the proposed method achieves higher accuracy than the conventional methods.

关 键 词：加权相机标定 边界提取 散焦模糊 模糊估计 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]