视场内无控制点的摄像机标定方法研究
发布时间：2018-10-05 21:29
【摘要】：摄像机标定是摄像测量的基础和关键步骤,摄像系统的微小误差在测量结果中可能被放大成千上万倍,所以要得到高精度的测量结果,必须对摄像系统进行高精度的标定。摄像机标定一般是通过空间位置精确已知的控制点与其对应的二维图像点之间的关系建立方程组,解算摄像机的内外参数。但是在海上或者空中以及其他视场内无法布置合作标志的场景下,没有控制点可以利用,以致现有的方法无法对摄像机进行标定,并且传统的标定方法在测量大范围远距离的目标时会因为模型和算法的原因引入较大误差。针对上述情况,本文分解成视场内无控制点的摄像机标定方法、依赖深度信息的畸变模型以及利用直线畸变特性去像差的新方法等三个问题进行解决。由于摄像机内参数的标定可以在实验室内高精度地完成,所以本文主要针对外参数的标定,并提出了两种标定方法:1)基于坐标系转换的标定方法,其基本思路是摄像机固定在刚性平台上,标定出摄像机相对于刚性平台的位置和姿态,然后将刚性平台整体移动到测量位置固定安装,标定出刚性平台相对于测量位置的世界坐标系的位置和姿态,然后通过坐标系之间的转换,解算摄像机在世界坐标系下的位置和姿态,即摄像机的外参数。2)基于激光直线的标定算法,其基本思路是通过激光在视场内创造若干条空间坐标已知的直线,根据空间直线和像直线的对应关系求解投影矩阵,从中分解摄像机的外参数。其基本做法是将激光发射器和全站仪镜筒固联在一起,激光发射器跟随全站仪镜筒转动,使得激光线在摄像机像面的不同位置成像,并根据全站仪此时的指向参数转换求解对应激光直线的空间三维信息,然后根据激光直线在空间中的位置和在像面上对应的位置之间的关系建立方程组,解算摄像机的外参数。本文针对用近距离的标定结果测量远距离控制点时会存在较大偏差的现象,在针孔模型的基础上,提出了一种像差系数依赖深度信息的畸变模型,并通过实验证明了该模型可以有效地提高测量的精度。此外,本文提出了一种利用直线畸变特性去像差的新方法:首先,利用直线畸变的对称性,先将直线畸变后的曲线校正成理想直线,再根据理想成像直线和实际畸变直线的像点之间对应的关系计算不同半径对应的畸变因子,然后根据这组数据拟合并比较各种径向畸变模型。本文为摄像机在视场内无控制点的情况下对大范围远距离测量的问题提供了有利的理论和技术支持,拓展了传统摄影测量的应用范围,有广阔的工程前景。
[Abstract]:Camera calibration is the basis and key step of camera measurement. The small error of camera system may be magnified by thousands of times in the measurement result. Therefore, in order to obtain the high precision measurement result, the camera system must be calibrated with high accuracy. Camera calibration is usually based on the relationship between the control points and the corresponding two-dimensional image points, and the internal and external parameters of the camera are solved. But in situations where cooperative signs cannot be placed at sea or in the air or in other fields of view, there are no control points to be used, so that the existing methods cannot calibrate the cameras. And the traditional calibration method will introduce a large error because of the model and algorithm when measuring the target in a wide range and long distance. In view of the above situation, this paper decomposes three problems: camera calibration method without control point in the field of view, distortion model dependent on depth information and a new method to remove aberration by using linear distortion characteristics. Since the calibration of camera parameters can be done in the laboratory with high precision, this paper mainly focuses on the calibration of external parameters, and proposes two calibration methods: 1) based on coordinate system transformation. The basic idea is that the camera is fixed on the rigid platform, the position and attitude of the camera relative to the rigid platform are calibrated, and then the rigid platform is moved to the measuring position and fixed installation. The position and attitude of the rigid platform in the world coordinate system relative to the measuring position are calibrated, and then the position and attitude of the camera in the world coordinate system are calculated by the transformation of the coordinate system. That is, the external parameter of the camera is based on the calibration algorithm of the laser straight line. The basic idea is to create some lines with known spatial coordinates through the laser in the field of view, and to solve the projection matrix according to the corresponding relation between the spatial line and the image line. Decomposes the external parameters of the camera. The basic method is to bind the laser transmitter and the total station tube together, and the laser emitter rotates with the total station mirror tube, so that the laser line can be imaged in different positions of the camera image plane. According to the transformation of the pointing parameters of the total station at this time, the spatial 3D information of the laser line is solved, then the equations are established according to the relationship between the position of the laser line in the space and the corresponding position on the image plane, and the external parameters of the camera are solved. Based on the pinhole model, a distortion model with aberration coefficient dependent on depth information is proposed in this paper, in view of the phenomenon that there is a large deviation in the measurement of remote control points by using the calibration results from close range, and based on the pinhole model, a distortion model with aberration coefficient dependent on depth information is proposed in this paper. The experimental results show that the model can effectively improve the accuracy of measurement. In addition, a new method of aberration is proposed. Firstly, by using the symmetry of linear distortion, the curve after distortion is corrected into an ideal line. Then the distortion factors corresponding to different radii are calculated according to the relationship between the image points of ideal imaging lines and actual distorted lines, and then the various radial distortion models are combined and compared according to the data set. This paper provides a favorable theoretical and technical support for the problem of large-scale long-distance measurement without control points in the field of view, expands the application of traditional photogrammetry, and has a broad engineering prospect.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TB853.1
，
本文编号：2254930
[Abstract]:Camera calibration is the basis and key step of camera measurement. The small error of camera system may be magnified by thousands of times in the measurement result. Therefore, in order to obtain the high precision measurement result, the camera system must be calibrated with high accuracy. Camera calibration is usually based on the relationship between the control points and the corresponding two-dimensional image points, and the internal and external parameters of the camera are solved. But in situations where cooperative signs cannot be placed at sea or in the air or in other fields of view, there are no control points to be used, so that the existing methods cannot calibrate the cameras. And the traditional calibration method will introduce a large error because of the model and algorithm when measuring the target in a wide range and long distance. In view of the above situation, this paper decomposes three problems: camera calibration method without control point in the field of view, distortion model dependent on depth information and a new method to remove aberration by using linear distortion characteristics. Since the calibration of camera parameters can be done in the laboratory with high precision, this paper mainly focuses on the calibration of external parameters, and proposes two calibration methods: 1) based on coordinate system transformation. The basic idea is that the camera is fixed on the rigid platform, the position and attitude of the camera relative to the rigid platform are calibrated, and then the rigid platform is moved to the measuring position and fixed installation. The position and attitude of the rigid platform in the world coordinate system relative to the measuring position are calibrated, and then the position and attitude of the camera in the world coordinate system are calculated by the transformation of the coordinate system. That is, the external parameter of the camera is based on the calibration algorithm of the laser straight line. The basic idea is to create some lines with known spatial coordinates through the laser in the field of view, and to solve the projection matrix according to the corresponding relation between the spatial line and the image line. Decomposes the external parameters of the camera. The basic method is to bind the laser transmitter and the total station tube together, and the laser emitter rotates with the total station mirror tube, so that the laser line can be imaged in different positions of the camera image plane. According to the transformation of the pointing parameters of the total station at this time, the spatial 3D information of the laser line is solved, then the equations are established according to the relationship between the position of the laser line in the space and the corresponding position on the image plane, and the external parameters of the camera are solved. Based on the pinhole model, a distortion model with aberration coefficient dependent on depth information is proposed in this paper, in view of the phenomenon that there is a large deviation in the measurement of remote control points by using the calibration results from close range, and based on the pinhole model, a distortion model with aberration coefficient dependent on depth information is proposed in this paper. The experimental results show that the model can effectively improve the accuracy of measurement. In addition, a new method of aberration is proposed. Firstly, by using the symmetry of linear distortion, the curve after distortion is corrected into an ideal line. Then the distortion factors corresponding to different radii are calculated according to the relationship between the image points of ideal imaging lines and actual distorted lines, and then the various radial distortion models are combined and compared according to the data set. This paper provides a favorable theoretical and technical support for the problem of large-scale long-distance measurement without control points in the field of view, expands the application of traditional photogrammetry, and has a broad engineering prospect.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TB853.1
，
本文编号：2254930
本文链接：https://www.wllwen.com/wenyilunwen/dianyingdianshilunwen/2254930.html
