基于合作标定物的相机标定方法研究

发布时间：2018-11-07 19:00

【摘要】：相机标定是计算机视觉领域的一个重要工作,尤其是视觉测量系统的一个重要的技术环节。相机标定的准确性对整个视觉测量系统的测量精度有很大的影响。一般而言,相机标定技术可以分为如下几种:视觉主动式标定技术、传统标定技术、相机本身标定技术。方法的选择往往基于实际的需求。对相机标定方法的要求是标定操作简单且精度高。基于此要求,本文的研究重点是传统方法,即基于合作标定物的标定技术。本文从相机标定基础理论出发,着重介绍了三维空间点到计算机帧存图像点的转换关系。为提出后续的相机标定方法提供了可靠的理论支撑。三种经典方法中的张正友平面标定方法因为其标定过程相对简单,标定结果较准确,在实际视觉测量系统中应用广泛。该方法本身只求解了两阶径向畸变系数,为了充分研究该方法的性能,通过仿真实验,探求了高阶径向畸变对该方法的影响。同时,本文在张氏标定方法的基础上提出了一种基于除法模型的相机标定方法,该方法巧妙地将畸变系数与单应性矩阵同时求解,使得在不使用非线性优化的情况下,得到与张氏方法精度相当的标定结果。同时该方法还考虑了畸变中心与主点不重合的情况,添加了畸变中心这一参数。通过仿真以及实验验证了该方法标定结果的准确性。基于合作标定物的相机标定方法中有三种经典的方法,其中线性方法是最简单最快捷的标定方法,但因其没有考虑相机畸变的存在,所以在标定精度上有所不足。本文在经典线性标定方法的基础上,根据实际工程需求,提出了一种基于标志器的相机标定方法。该方法是在线性标定方法的基础上,继续估计径向畸变,利用非线性优化对所有参数进行优化。通过实际实验证明了该方法的正确性。最后给出了本文所搭建的相机标定系统,并进行了10次完整的相机标定,得到10个标定结果,通过标定结果的分析,证明该系统简单、实用、功能正确。
[Abstract]:Camera calibration is an important work in the field of computer vision, especially in the vision measurement system. The accuracy of camera calibration has great influence on the measurement accuracy of the whole vision measurement system. Generally speaking, camera calibration technology can be divided into the following: visual active calibration technology, traditional calibration technology, camera itself calibration technology. The choice of methods is often based on actual requirements. The requirement of camera calibration method is that the calibration operation is simple and the precision is high. Based on this requirement, this paper focuses on the traditional method, that is, the calibration technology based on cooperative calibration object. Based on the basic theory of camera calibration, this paper mainly introduces the conversion relationship between three dimensional space points and computer frame memory image points. It provides a reliable theoretical support for the subsequent camera calibration method. Because the calibration process is relatively simple and the calibration results are more accurate, the three classical methods are widely used in the practical vision measurement system. In order to fully study the performance of the method, the influence of higher order radial distortion on the method is investigated by simulation experiments. At the same time, on the basis of Zhang's calibration method, a camera calibration method based on division model is proposed in this paper. In this method, the distortion coefficient is solved simultaneously with the monoclinic matrix, so that the nonlinear optimization is not used. The calibration results are equivalent to those of Zhang's method. At the same time, the distortion center is not coincident with the main point, and the parameter of distortion center is added. The accuracy of the calibration results is verified by simulation and experiments. There are three classical methods for camera calibration based on cooperative calibration object, among which linear method is the simplest and quickest method, but because it does not consider the existence of camera distortion, it has some shortcomings in calibration accuracy. Based on the classical linear calibration method and the practical engineering requirements, a camera calibration method based on marker is proposed in this paper. Based on the linear calibration method, the radial distortion is estimated and all parameters are optimized by nonlinear optimization. The correctness of the method is proved by practical experiments. Finally, the camera calibration system in this paper is presented, and 10 complete camera calibration results are obtained. Through the analysis of calibration results, it is proved that the system is simple, practical and functional.
【学位授予单位】：中国科学院研究生院(光电技术研究所)
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.41

【参考文献】