WO2020118637A1 - 摄像机标定板、标定方法以及摄像机 - Google Patents

摄像机标定板、标定方法以及摄像机 Download PDF

Info

Publication number
WO2020118637A1
WO2020118637A1 PCT/CN2018/120963 CN2018120963W WO2020118637A1 WO 2020118637 A1 WO2020118637 A1 WO 2020118637A1 CN 2018120963 W CN2018120963 W CN 2018120963W WO 2020118637 A1 WO2020118637 A1 WO 2020118637A1
Authority
WO
WIPO (PCT)
Prior art keywords
calibration
camera
dot
dots
image
Prior art date
Application number
PCT/CN2018/120963
Other languages
English (en)
French (fr)
Inventor
陈辉
邱文胜
陈鹏宇
Original Assignee
昂纳自动化技术(深圳)有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 昂纳自动化技术(深圳)有限公司 filed Critical 昂纳自动化技术(深圳)有限公司
Priority to PCT/CN2018/120963 priority Critical patent/WO2020118637A1/zh
Publication of WO2020118637A1 publication Critical patent/WO2020118637A1/zh

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/80Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration

Definitions

  • the present invention relates to the field of camera calibration, and more particularly, to a camera calibration board, calibration method, and camera.
  • the calibration of camera parameters is a crucial link, especially in vision measurement applications, the calibration accuracy of which directly affects the accuracy of the entire measurement system.
  • the camera calibration requires the use of a special structure calibration plate to extract precise point and line features.
  • the checkerboard calibration board obtains point features by locating the checkerboard corners. In order to ensure the accuracy of corner extraction, high imaging quality is usually required.
  • the dot-shaped calibration plate obtains point features by locating the center of the dot, and can obtain high-precision position information through sub-pixel edges, ellipse fitting and other methods.
  • the calibration board can be divided into two types, with or without reference, according to whether it has a reference or not.
  • Method 1 OpenCV calibration board and corresponding feature detection algorithm.
  • the OpenCV calibration board belongs to the calibration board without a reference checkerboard, as shown in Figure 1.
  • the calibration board is composed of checkerboards with a certain number of rows and columns, and the corner points of the checkerboard are the characteristic information of the calibration board.
  • the general detection steps are: (1) corner detection; (2) sorting in the order specified in advance.
  • the advantages of the OpenCV calibration board and the corresponding feature detection algorithm are that the calibration board has a simple structure and the algorithm is simple and easy to implement; the disadvantage is that the calibration board does not have a reference, and it is necessary to manually specify the reference, which is inconvenient to use; and the characteristic points of the calibration board are corners, using grayscale or Edge information positioning, positioning accuracy is not high.
  • Mode 2 Halcon calibration board and corresponding feature detection algorithm.
  • the Halcon calibration plate is a calibration plate with a reference dot, as shown in Figure 2.
  • the calibration plate is composed of a number of dots in a specific row and column and the outer border of the dot.
  • the center of the dot is the characteristic information of the calibration plate.
  • the small triangle on the border contains the orientation information of the calibration plate.
  • the general detection steps are: (1) detection of the calibration board area; (2) detection of the calibration board circle point, that is, edge detection + ellipse fitting; (3) calculation of the reference point using the area, and sorting of the circle point coordinates according to the reference point.
  • the advantages of the Halcon calibration board and the corresponding feature detection algorithm are that the calibration board has a reference and it is easy to use;
  • the characteristic point of the fixed plate is a circular mark point, which is positioned by the ellipse fitting center, and the positioning accuracy is high.
  • the disadvantage is that the positioning datum is on the frame, and the calibration plate needs to be completely in the field of view; and the corresponding algorithm requires that the mark point needs to be detected completely, and it cannot adapt to the situation of defects or noise points, and the robustness is not good.
  • the technical problem to be solved by the present invention is to provide a camera calibration board, a calibration method, and a camera in view of the above-mentioned defects of the prior art.
  • the technical solution adopted by the present invention to solve its technical problem is: constructing a camera calibration plate, including a plurality of dots, the dots are divided into a reference dot and a reference dot;
  • the reference dots and the reference dots are arranged in a checkerboard shape, all rows are parallel, all columns are parallel, all rows and all columns are perpendicular to each other; wherein the spacing of adjacent dots in each row is equal The distance between adjacent dots in each column is equal, and the distance between adjacent dots in each row is equal to the distance between adjacent dots in each column.
  • the reference dots are arranged to form a rectangular coordinate system, wherein one of the reference dots is used as the origin of the rectangular coordinate system, and one reference is adjacent to the origin
  • the round point serves as the x-axis of the rectangular coordinate system
  • the reference round point adjacent to the origin serves as the y-axis of the rectangular coordinate system
  • a reference circle point is set as a direction calibration point on the negative half axis of the X axis, or a reference circle point is set as a direction calibration point on the negative half axis of the y axis.
  • the distance between adjacent dots in each row is d, 1J
  • the present invention also provides a camera calibration method, the camera is calibrated using the camera calibration board as described above, the method includes:
  • edge extraction is performed on the calibration board image collected by the camera to obtain the edge contour of each dot
  • the camera calibration method according to the present invention before performing edge extraction on the calibration plate image collected by the camera, further includes:
  • the camera acquires a calibration board image, and performs image quality preprocessing on the calibration board image.
  • the edge extraction of the calibration plate image collected by the camera includes:
  • the edge extraction of the calibration plate image collected by the camera includes:
  • the ellipse fitting of the edge contour of each dot includes:
  • a robust ellipse fitting algorithm is used to perform ellipse fitting on the edge contour of each dot.
  • the sorting of all marker points is completed according to the reference dot, the coordinate system, and the parameters of the calibration board to obtain a one-to-one correspondence between the image coordinates and the world Coordinates include:
  • a growth-based mark point sorting algorithm is used to determine the orientation of all mark points in the reference coordinate system, and a one-to-one corresponding image coordinate and world coordinate are obtained by combining the parameters of the calibration plate.
  • the completion of the calibration calculation of the camera parameters according to the image coordinates and the world coordinates includes:
  • the present invention also provides a camera, the camera includes a processor, the processor is used to execute the computer program stored in the memory to implement the camera calibration method as described above.
  • a camera calibration board, calibration method and camera implementing the present invention have the following beneficial effects:
  • the present invention uses circular marker points of different sizes, and uses the reference circle point as a reference for detection.
  • the calibration method includes: extracting the edge of the calibration board image collected by the camera to obtain the edge contour of each dot; ellipse fitting the edge contour of each dot to obtain the fitting ellipse corresponding to each dot; according to Fit the relationship between the size and position of the ellipse to locate the reference ellipse; use the reference ellipse as the seed point to complete the sorting of all marker points to obtain one-to-one correspondence between image coordinates and world coordinates; use image coordinates and world coordinates to complete the camera Calibration calculation of parameters.
  • the calibration plate is placed flexibly, as long as the reference round point is in the field of view, and the point spacing is provided, and the number of rows and columns of the calibration plate marking points is not required, the camera calibration can be completed, effectively improving the calibration Robustness and accuracy.
  • FIG. 1 is a schematic structural diagram of a checkerboard calibration plate in the prior art
  • FIG. 2 is a schematic structural diagram of a dot-shaped calibration plate in the prior art
  • FIG. 3 is a schematic structural view of a camera calibration board of the present invention.
  • FIG. 4 is a flowchart of a camera calibration method of the present invention.
  • FIG. 5 is a schematic structural view of a camera of the present invention.
  • the camera calibration plate provided in this embodiment includes a plurality of dots, where the dots are solid dots, that is, the dots are filled with colors, for example, the dots are filled with black.
  • the dots are divided into reference dots and reference dots; alternatively, the reference dots in this embodiment are larger than the reference dots, that is, the diameter of the reference dots is larger than the diameter of the reference dots.
  • the reference dots and reference dots are arranged in a checkerboard pattern, all rows are parallel, all columns are parallel, all rows and all columns are perpendicular to each other; wherein the spacing of adjacent dots in each row is equal, adjacent in each column
  • the pitch of dots is equal, and the pitch of adjacent dots in each row is equal to the pitch of adjacent dots in each column.
  • the spacing between adjacent rows mentioned here refers to the distance between the centers of two adjacent dots on the same column, and the spacing between adjacent columns refers to two adjacent circles on the same row The distance between the centers of points.
  • the reference dots in the camera calibration plate of this embodiment are arranged to form a rectangular coordinate system, wherein a reference dot is used as the origin of the rectangular coordinate system, where the origin refers to the center of the reference dot; it is relative to the origin
  • the adjacent reference point is the x-axis of the rectangular coordinate system, that is, the line between the origin and the center of the reference point is the x- axis; the reference point adjacent to the origin is the y-axis of the rectangular coordinate system, that is, the origin and
  • the straight line where the center of the reference dot is located is the y-axis.
  • the distance between adjacent dots in each row in the camera calibration plate of this embodiment is d, then
  • the distance from the boundary to the outermost dot is required to be greater than or equal to d, the greater the blanking distance between the boundary and the dot, the better the anti-interference performance.
  • the calibration plate includes m rows and n columns, where m and n are integers greater than 1, and 1J
  • the calibration plate of this embodiment uses several large circular mark points as a reference to facilitate reference detection, and only needs to use the same algorithm as the ordinary feature points.
  • the calibration plate reference can adapt to tilt, flip, mirror, etc., and can uniquely determine the orientation of the calibration plate.
  • the above calibration board reference is located at the center of the calibration board, and the mark points on the edge of the calibration board are outside the image field of view, which does not affect the detection of the reference, which provides early conditions for the subsequent algorithm to adapt to the disadvantages and noises.
  • the camera calibration method of this embodiment uses the camera calibration board of the foregoing embodiment, and the method includes:
  • S1 Perform edge extraction on the calibration plate image collected by the camera to obtain the edge contour of each dot.
  • Sobel algorithm, Canny edge detection operator, etc. may be used to extract continuous edge contour points in the calibration plate image.
  • the edge extraction of the calibration board image collected by the camera includes: performing edge detection on the calibration board image to obtain a set of points of the edge contour, and the abnormal edge cannot be the edge of the mark point, and the abnormal edge in the edge contour needs to be screened out .
  • conditions such as gradient threshold, number of contour points, and closing conditions can be used to preliminarily filter a portion of the edges that are not likely to be landmarks.
  • performing edge extraction on the calibration board image collected by the camera includes: first performing pixel-level edge extraction on the calibration board image, and then performing sub-pixel positioning on the obtained edge.
  • edge sub-pixel positioning algorithms are: interpolation-based sub-pixel edge detection algorithm, fitting-based sub-pixel edge detection algorithm, moment-based sub-pixel edge detection algorithm, etc., which can be selected according to the actual situation, or choose other edge sub-pixel positioning algorithms with the same effect as this embodiment.
  • the ellipse fitting algorithm commonly used for the ellipse fitting of the edge contour of each dot is the least squares method. It may be considered to use the algebraic fitting method and the geometric fitting method to fit a certain error, size, shape Equally effective ellipse. If there are many contour noise points extracted from the above-mentioned edges, a robust ellipse fitting method may be further considered, that is, a robust ellipse fitting algorithm is used to perform ellipse fitting on the edge contour of each dot to improve the accuracy of ellipse fitting. [0058] S3. Compare the size and position relationship between the fitted ellipses to determine the reference ellipse and the coordinate system composed of the reference ellipse, where the reference ellipse corresponds to the reference circle point.
  • the size of the reference circle point in the camera calibration plate used in this embodiment is larger than the reference circle point, that is, the fitting ellipse corresponding to the reference circle point is larger than the fitting ellipse corresponding to the reference circle point, and the fitting ellipse is compared by comparison The size between them, combined with the positional relationship between the fitted ellipses, determines the reference ellipse, where the positional relation between the fitted ellipses corresponds to the positional relation between the reference dot and the reference dot in the calibration plate. Then, the coordinate system composed of the reference ellipse is determined in combination with the preset coordinate system composition rules.
  • one reference circle point is used as the origin of the rectangular coordinate system, where the origin point refers to the center of the reference circle point;
  • a reference circle point adjacent to the origin is used as the x-axis of the rectangular coordinate system, that is, the line between the origin and the center of the reference circle is the x axis;
  • a reference circle point adjacent to the origin is the y axis of the rectangular coordinate system, That is, the straight line on which the origin and the center of the reference dot are located is the y-axis.
  • reference dots dot 1, dot 2, dot 3, dot 4, where the reference dot 1 is the origin of the rectangular coordinate system, the reference dot 1 and the reference dot 2 is the x-axis of the rectangular coordinate system, the reference circle 1 and the reference circle 3 are the y-axis of the rectangular coordinate system, and the reference circle 4 is the direction calibration point on the negative half-axis of the x-axis.
  • the reference mark point is used as a seed point
  • a growth-based mark point sorting algorithm is used to determine the orientation of all mark points in the reference coordinate system
  • a one-to-one correspondence between image coordinates and world coordinates is obtained in combination with the parameters of the calibration plate.
  • the calibration plate is placed flexibly, as long as the reference dots are in the field of view, and the point spacing is provided, and the number of rows and columns of the calibration plate marking points is not required, the camera calibration can be completed, effectively improving Robustness and accuracy of calibration.
  • the camera calibration method of this embodiment before performing edge extraction on the calibration board image collected by the camera further includes: the camera acquiring the calibration board image, and performing image quality preprocessing on the calibration board image, the image Quality preprocessing includes image filtering, image enhancement, etc.
  • the preprocessed image can improve the accuracy of subsequent edge processing.
  • the image quality is improved by preprocessing the calibration plate image, so that the accuracy of subsequent edge processing can be improved.
  • this embodiment provides a camera.
  • the camera includes a processor, and the processor is used to implement the camera calibration method as described above when executing a computer program stored in a memory.
  • the calibration plate is placed flexibly, as long as the reference round point is in the field of view, providing the point spacing, without providing the number of rows and columns of the calibration plate marking points, the calibration of the camera can be completed , Effectively improve the calibration robustness and accuracy.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Length Measuring Devices By Optical Means (AREA)

Abstract

本发明涉及一种摄像机标定板、标定方法以及摄像机。本发明采用不同尺寸的圆形标志点,以基准圆点作为基准进行检测。该标定方法包括:对摄像机采集的标定板图像进行边缘提取,得到每个圆点的边缘轮廓;对每个圆点的边缘轮廓进行椭圆拟合,得到每个圆点对应的拟合椭圆;根据拟合椭圆之间的大小与位置关系,定位出基准椭圆;以基准椭圆为种子点,完成所有标志点的排序,得到一一对应的图像坐标与世界坐标;利用图像坐标与世界坐标,完成摄像机参数的标定计算。通过实施本发明,在标定过程中,标定板放置灵活,只需保证基准圆点位于视野内,提供点间距,不需要提供标定板标志点的行列数,即可完成摄像机的标定,有效提升标定的鲁棒性与精度。

Description

摄像机标定板、 标定方法以及摄像机 技术领域
[0001] 本发明涉及摄像机标定领域, 更具体地说, 涉及一种摄像机标定板、 标定方法 以及摄像机。
背景技术
[0002] 在机器视觉应用中, 摄像机参数的标定是一个至关重要的环节, 尤其是视觉测 量应用中, 其标定精度直接影响了整个测量系统的精度。 摄像机标定需要用到 特殊结构的标定板, 从而提取精确的点线特征。 目前, 5见有技术中标定板主要 分两种: 棋盘格标定板与圆点型标定板。 其中棋盘格标定板通过定位棋盘格角 点来获取点特征, 为了确保角点的提取精度, 通常要求较高的成像质量。 圆点 型标定板通过定位圆点中心来获取点特征, 通过亚像素边缘、 椭圆拟合等方法 , 可以得到高精度的位置信息。 此外, 标定板根据是否带有基准可以分为带基 准、 不带基准两种。
[0003] 进一步, 5见有技术通常采用以前两种方式进行标定:
[0004] 方式一: OpenCV标定板及相应特征检测算法。 OpenCV标定板属于不带基准棋 盘格标定板, 如图 1所示。 该标定板由特定行列数量的棋盘格组成, 棋盘格角点 即为该标定板的特征信息。 一般检测步骤为: (1) 角点检测; (2) 按照事先 指定顺序排序。 OpenCV标定板及相应特征检测算法的优点是标定板结构简单, 算法简单易实现; 缺点是标定板不带基准, 需要手动指定基准, 使用不便; 且 标定板特征点为角点, 采用灰度或边缘信息定位, 定位精度不高。
[0005] 方式二: Halcon标定板及相应特征检测算法。 Halcon标定板属于带基准圆点型 标定板, 如图 2所示。 该标定板由特定行列数量的圆点与圆点的外边框组成, 圆 点的中心即为该标定板的特征信息, 边框上的小三角包含了该标定板的方位信 息。 一般检测步骤为: (1) 标定板区域检测; (2) 标定板圆点检测, 即边缘 检测 +椭圆拟合; (3) 利用区域计算出基准点, 按照基准点对圆点坐标进行排 序。 Halcon标定板及相应特征检测算法的优点是标定板带基准, 使用方便; 且标 定板特征点为圆形标志点, 采用椭圆拟合中心定位, 定位精度高。 缺点是定位 基准在边框上, 标定板需要完全位于视野内; 且相应算法要求标志点需要刚好 完全检测出, 不能适应缺点或者杂点的情况, 鲁棒性不好。
发明概述
技术问题
[0006] 本发明要解决的技术问题在于, 针对现有技术的上述缺陷, 提供一种摄像机标 定板、 标定方法以及摄像机。
问题的解决方案
技术解决方案
[0007] 本发明解决其技术问题所采用的技术方案是: 构造一种摄像机标定板, 包括多 个圆点, 所述圆点分为基准圆点和参考圆点;
[0008] 所述基准圆点和所述参考圆点呈棋盘状排列, 所有行之间平行, 所有列之间平 行, 所有行和所有列相互垂直; 其中每行中相邻圆点的间距相等, 每列中相邻 圆点的间距相等, 且每行中相邻圆点的间距等于每列中相邻圆点的间距。
[0009] 进一步, 本发明所述的摄像机标定板, 所述基准圆点排列形成直角坐标系, 其 中一所述基准圆点作为直角坐标系的原点, 与所述原点相邻的一所述基准圆点 作为直角坐标系的 x轴, 与所述原点相邻的一所述基准圆点作为直角坐标系的 y 轴;
[0010] 在所述 X轴的负半轴上设置一所述基准圆点作为方向标定点, 或在所述 y轴的负 半轴上设置一所述基准圆点作为方向标定点。
[0011] 进一步, 本发明所述的摄像机标定板, 每行中相邻圆点的间距为 d, 贝 1J
[0012] 所述基准圆点的直径 d ^d/2;
[0013] 所述参考圆点的直径 d s=d/4
[0014] 另, 本发明还提供一种摄像机标定方法, 所述摄像机使用如上述的摄像机标定 板进行标定, 所述方法包括:
[0015] 对所述摄像机采集的标定板图像进行边缘提取, 得到每个圆点的边缘轮廓;
[0016] 对每个圆点的边缘轮廓进行椭圆拟合, 得到每个圆点对应的拟合椭圆;
[0017] 比较所述拟合椭圆之间的大小与位置关系确定基准椭圆以及由所述基准椭圆组 成的坐标系, 其中所述基准椭圆对应基准圆点;
[0018] 根据所述基准圆点、 坐标系、 以及所述标定板的参数完成对所有标志点的排序 , 得到一一对应的图像坐标与世界坐标;
[0019] 根据所述图像坐标与世界坐标, 完成摄像机参数的标定计算。
[0020] 进一步, 本发明所述的摄像机标定方法, 在所述对所述摄像机采集的标定板图 像进行边缘提取之前还包括:
[0021] 所述摄像机获取标定板图像, 对所述标定板图像进行图像质量预处理。
[0022] 进一步, 本发明所述的摄像机标定方法, 所述对所述摄像机采集的标定板图像 进行边缘提取包括:
[0023] 对所述标定板图像进行边缘检测, 得到边缘轮廓的点集合, 筛选剔除掉所述边 缘轮廓中的异常边缘。
[0024] 进一步, 本发明所述的摄像机标定方法, 所述对所述摄像机采集的标定板图像 进行边缘提取包括:
[0025] 首先对所述标定板图像进行像素级的边缘提取, 再对得到的边缘进行亚像素级 定位。
[0026] 进一步, 本发明所述的摄像机标定方法, 所述对每个圆点的边缘轮廓进行椭圆 拟合包括:
[0027] 使用鲁棒的椭圆拟合算法对每个圆点的边缘轮廓进行椭圆拟合。
[0028] 进一步, 本发明所述的摄像机标定方法, 所述根据所述基准圆点、 坐标系、 以 及所述标定板的参数完成对所有标志点的排序, 得到一一对应的图像坐标与世 界坐标包括:
[0029] 以基准标志点作为种子点, 采用基于生长的标志点排序算法, 确定基准坐标系 下所有标志点的方位, 结合所述标定板的参数得到一一对应的图像坐标与世界 坐标。
[0030] 进一步, 本发明所述的摄像机标定方法, 所述根据所述图像坐标与世界坐标, 完成摄像机参数的标定计算包括:
[0031] 根据所述图像坐标与世界坐标, 计算出理想状态下无畸变的摄像机参数, 作为 摄像机参数初值; [0032] 通过最优化方法最小化重投影误差, 迭代求出精确的摄像机参数, 完成对所述 摄像机的参数标定。
[0033] 另, 本发明还提供一种摄像机, 所述摄像机包括处理器, 所述处理器用于执行 存储器中存储的计算机程序时实现如上述的摄像机标定方法。
发明的有益效果
有益效果
[0034] 实施本发明的一种摄像机标定板、 标定方法以及摄像机, 具有以下有益效果: 本发明采用不同尺寸的圆形标志点, 以基准圆点作为基准进行检测。 该标定方 法包括: 对摄像机采集的标定板图像进行边缘提取, 得到每个圆点的边缘轮廓 ; 对每个圆点的边缘轮廓进行椭圆拟合, 得到每个圆点对应的拟合椭圆; 根据 拟合椭圆之间的大小与位置关系, 定位出基准椭圆; 以基准椭圆为种子点, 完 成所有标志点的排序, 得到一一对应的图像坐标与世界坐标; 利用图像坐标与 世界坐标, 完成摄像机参数的标定计算。 通过实施本发明, 在标定过程中, 标 定板放置灵活, 只需保证基准圆点位于视野内, 提供点间距, 不需要提供标定 板标志点的行列数, 即可完成摄像机的标定, 有效提升标定的鲁棒性与精度。 对附图的简要说明
附图说明
[0035] 下面将结合附图及实施例对本发明作进一步说明, 附图中:
[0036] 图 1是现有技术中棋盘格标定板的结构示意图;
[0037] 图 2是现有技术中圆点型标定板的结构示意图;
[0038] 图 3是本发明一种摄像机标定板的结构示意图;
[0039] 图 4是本发明一种摄像机标定方法流程图;
[0040] 图 5是本发明一种摄像机的结构示意图。
实施该发明的最佳实施例
本发明的最佳实施方式
[0041] 为了对本发明的技术特征、 目的和效果有更加清楚的理解, 现对照附图详细说 明本发明的具体实施方式。 发明实施例
实施例
[0042] 如图 3所示, 本实施例提供的像机标定板包括多个圆点, 这里的圆点为实心圆 点, 即圆点内填充颜色, 例如圆点中填充黑色。 圆点分为基准圆点和参考圆点 ; 作为选择, 本实施例中的基准圆点大于参考圆点, 即基准圆点的直径大于参 考圆点的直径。 其中基准圆点和参考圆点呈棋盘状排列, 所有行之间平行, 所 有列之间平行, 所有行和所有列相互垂直; 其中每行中相邻圆点的间距相等, 每列中相邻圆点的间距相等, 且每行中相邻圆点的间距等于每列中相邻圆点的 间距。 这里所说的相邻行之间的间距指的是同一列上的两个相邻圆点的圆心之 间的距离, 相邻列之间的间距指的是同一行上的两个相邻圆点的圆心之间的距 离。
[0043] 进一步, 本实施例的摄像机标定板中基准圆点排列形成直角坐标系, 其中一基 准圆点作为直角坐标系的原点, 这里的原点指的是该基准圆点的圆心; 与原点 相邻的一基准圆点作为直角坐标系的 x轴, 即原点与该基准圆点的圆心所在的直 线为 x轴; 与原点相邻的一基准圆点作为直角坐标系的 y轴, 即原点与该基准圆 点的圆心所在的直线为 y轴。 在 x轴的负半轴上设置一基准圆点作为方向标定点 , 或在 y轴的负半轴上设置一基准圆点作为方向标定点。 例如图 3中所示, 设置 4 个基准圆点: 圆点 1、 圆点 2、 圆点 3、 圆点 4, 其中基准圆点 1作为直角坐标系的 原点, 基准圆点 1和基准圆点 2作为直角坐标系的 x轴, 基准圆点 1和基准圆点 3作 为直角坐标系的 y轴, 基准圆点 4在 x轴的负半轴上作为方向标定点。
[0044] 作为选择, 本实施例的摄像机标定板中每行中相邻圆点的间距为 d, 则
[0045] 基准圆点的直径 d 1=d/2;
[0046] 参考圆点的直径 d s=d/4。
[0047] 作为选择, 若摄像机标定板的四周设置有边界, 要求边界距离最外侧圆点的距 离大于等于 d, 边界与圆点之间的留白距离越大, 抗干扰性越好。 设标定板包括 m行和 n列, 其中 m、 n为大于 1的整数, 贝 1J
[0048] 标定板的宽度
Figure imgf000007_0001
(n+1) d;
[0049] 标定板的高 (m+1) d; [0050] 需要说明的是, 本实施例的标定板尺寸只是其中一种实施方式, 用于说明标定 板中基准圆点和参考圆点的大小和排布规律, 并不用于限定本发明的保护范围
[0051] 本实施例的标定板用较大的几个圆形标志点作为基准, 方便进行基准检测, 只 需采用与普通特征点相同的算法即可。 上述标定板基准能适应倾斜、 翻转、 镜 像等情况, 能唯一确定标定板方位。 上述标定板基准位于标定板中心处, 标定 板边缘处标志点在图像视野范围外也不影响基准的检测, 为后续算法适应缺点 、 杂点的情况提供先期条件。
实施例
[0052] 如图 4所示, 本实施例的摄像机标定方法使用如上述实施例的摄像机标定板, 该方法包括:
[0053] S1、 对摄像机采集的标定板图像进行边缘提取, 得到每个圆点的边缘轮廓。
[0054] 具体的, 可使用 Sobel算法、 Canny边缘检测算子等, 提取出标定板图像中连续 的边缘轮廓点。 作为选择, 对摄像机采集的标定板图像进行边缘提取包括: 对 标定板图像进行边缘检测, 得到边缘轮廓的点集合, 异常边缘不可能是标志点 的边缘, 需筛选剔除掉边缘轮廓中的异常边缘。 作为选择, 可利用梯度阈值、 轮廓点数、 闭合情况等条件可初步过滤一部分不可能是标志点的边缘。
[0055] 进一步, 对摄像机采集的标定板图像进行边缘提取包括: 首先对标定板图像进 行像素级的边缘提取, 再对得到的边缘进行亚像素级定位。 作为选择, 常用的 边缘亚像素定位算法有: 基于插值的亚像素边缘检测算法、 基于拟合的亚像素 边缘检测算法、 基于矩特征的亚像素边缘检测算法等, 可根据实际情况选用, 也可选择与本实施例作用相同的其他边缘亚像素定位算法。
[0056] S2、 对每个圆点的边缘轮廓进行椭圆拟合, 得到每个圆点对应的拟合椭圆。
[0057] 具体的, 对每个圆点的边缘轮廓进行椭圆拟合常用的椭圆拟合算法为最小二乘 法, 可考虑选用代数拟合法与几何拟合法, 拟合出满足一定误差、 尺寸、 形状 等条件的有效椭圆。 若上述边缘提取的轮廓杂点较多, 可以进一步考虑采用鲁 棒的椭圆拟合法, 即使用鲁棒的椭圆拟合算法对每个圆点的边缘轮廓进行椭圆 拟合, 提高椭圆拟合精度。 [0058] S3、 比较拟合椭圆之间的大小与位置关系确定基准椭圆以及由基准椭圆组成的 坐标系, 其中基准椭圆对应基准圆点。
[0059] 具体的, 本实施例中使用的摄像机标定板中基准圆点的尺寸大于参考圆点, 即 基准圆点对应的拟合椭圆大于参考圆点对应的拟合椭圆, 通过比较拟合椭圆之 间的大小, 再结合拟合椭圆之间的位置关系确定基准椭圆, 其中拟合椭圆之间 的位置关系与标定板中基准圆点和参考圆点的位置关系相对应。 然后结合预设 的坐标系构成规则确定由基准椭圆组成的坐标系, 在建立坐标系过程中, 其中 一基准圆点作为直角坐标系的原点, 这里的原点指的是该基准圆点的圆心; 与 原点相邻的一基准圆点作为直角坐标系的 x轴, 即原点与该基准圆点的圆心所在 的直线为 x轴; 与原点相邻的一基准圆点作为直角坐标系的 y轴, 即原点与该基 准圆点的圆心所在的直线为 y轴。 在 x轴的负半轴上设置一基准圆点作为方向标 定点, 或在 y轴的负半轴上设置一基准圆点作为方向标定点。 例如图 3中所示, 设置 4个基准圆点: 圆点 1、 圆点 2、 圆点 3、 圆点 4, 其中基准圆点 1作为直角坐 标系的原点, 基准圆点 1和基准圆点 2作为直角坐标系的 x轴, 基准圆点 1和基准 圆点 3作为直角坐标系的 y轴, 基准圆点 4在 x轴的负半轴上作为方向标定点。
[0060] S4、 根据基准圆点、 坐标系、 以及标定板的参数完成对所有标志点的排序, 得 到 -对应的图像坐标与世界坐标。
[0061] 具体的, 以基准标志点作为种子点, 采用基于生长的标志点排序算法, 确定基 准坐标系下所有标志点的方位, 结合标定板的参数得到一一对应的图像坐标与 世界坐标。
[0062] S5、 根据图像坐标与世界坐标, 完成摄像机参数的标定计算。
[0063] 具体的, 根据图像坐标与世界坐标, 计算出理想状态下无畸变的摄像机参数, 作为摄像机参数初值; 通过最优化方法最小化重投影误差, 迭代求出精确的摄 像机参数, 完成对摄像机的参数标定。
[0064] 本实施例在标定过程中, 标定板放置灵活, 只需保证基准圆点位于视野内, 提 供点间距, 不需要提供标定板标志点的行列数, 即可完成摄像机的标定, 有效 提升标定的鲁棒性与精度。
实施例 [0065] 在上述实施例的基础上, 本实施例的摄像机标定方法在对摄像机采集的标定板 图像进行边缘提取之前还包括: 摄像机获取标定板图像, 对标定板图像进行图 像质量预处理, 图像质量预处理包括图像滤波、 图像增强等, 预处理后的图像 可提高后续边缘处理的准确度。
[0066] 本实施例中通过对标定板图像的预处理提高图像质量, 从而可提高后续边缘处 理的准确度。
实施例
[0067] 如图 5所示, 本实施例提供一种摄像机, 摄像机包括处理器, 处理器用于执行 存储器中存储的计算机程序时实现如上述的摄像机标定方法。
[0068] 通过实施本实施例, 在标定过程中, 标定板放置灵活, 只需保证基准圆点位于 视野内, 提供点间距, 不需要提供标定板标志点的行列数, 即可完成摄像机的 标定, 有效提升标定的鲁棒性与精度。
[0069] 本说明书中各个实施例采用递进的方式描述, 每个实施例重点说明的都是与其 他实施例的不同之处, 各个实施例之间相同相似部分互相参见即可。 对于实施 例公开的装置而言, 由于其与实施例公开的方法相对应, 所以描述的比较简单 , 相关之处参见方法部分说明即可。
[0070] 专业人员还可以进一步意识到, 结合本文中所公开的实施例描述的各示例的单 元及算法步骤, 能够以电子硬件、 计算机软件或者二者的结合来实现, 为了清 楚地说明硬件和软件的可互换性, 在上述说明中已经按照功能一般性地描述了 各示例的组成及步骤。 这些功能究竟以硬件还是软件方式来执行, 取决于技术 方案的特定应用和设计约束条件。 专业技术人员可以对每个特定的应用来使用 不同方法来实现所描述的功能, 但是这种实现不应认为超出本发明的范围。
[0071] 结合本文中所公开的实施例描述的方法或算法的步骤可以直接用硬件、 处理器 执行的软件模块, 或者二者的结合来实施。 软件模块可以置于随机存储器 (RA M) 、 内存、 只读存储器 (ROM) 、 电可编程 ROM、 电可擦除可编程 ROM、 寄 存器、 硬盘、 可移动磁盘、 CD-ROM、 或技术领域内所公知的任意其它形式的 存储介质中。
[0072] 以上实施例只为说明本发明的技术构思及特点, 其目的在于让熟悉此项技术的 人士能够了解本发明的内容并据此实施, 并不能限制本发明的保护范围。 凡跟 本发明权利要求范围所做的均等变化与修饰, 均应属于本发明权利要求的涵盖 范围。

Claims

权利要求书
[权利要求 1] 一种摄像机标定板, 其特征在于, 包括多个圆点, 所述圆点分为基准 圆点和参考圆点;
所述基准圆点和所述参考圆点呈棋盘状排列, 所有行之间平行, 所有 列之间平行, 所有行和所有列相互垂直; 其中每行中相邻圆点的间距 相等, 每列中相邻圆点的间距相等, 且每行中相邻圆点的间距等于每 列中相邻圆点的间距。
[权利要求 2] 根据权利要求 1所述的摄像机标定板, 其特征在于, 所述基准圆点排 列形成直角坐标系, 其中一所述基准圆点作为直角坐标系的原点, 与 所述原点相邻的一所述基准圆点作为直角坐标系的 x轴, 与所述原点 相邻的一所述基准圆点作为直角坐标系的 y轴;
在所述 x轴的负半轴上设置一所述基准圆点作为方向标定点, 或在所 述 y轴的负半轴上设置一所述基准圆点作为方向标定点。
[权利要求 3] 根据权利要求 1所述的摄像机标定板, 其特征在于, 每行中相邻圆点 的间距为 d, 贝 1J
所述基准圆点的直径 d 1=d/2;
所述参考圆点的直径 d s=d/4。
[权利要求 4] 一种摄像机标定方法, 其特征在于, 所述摄像机使用如权利要求 1至 3 任一项所述的摄像机标定板进行标定, 所述方法包括:
对所述摄像机采集的标定板图像进行边缘提取, 得到每个圆点的边缘 轮廓;
对每个圆点的边缘轮廓进行椭圆拟合, 得到每个圆点对应的拟合椭圆 比较所述拟合椭圆之间的大小与位置关系确定基准椭圆以及由所述基 准椭圆组成的坐标系, 其中所述基准椭圆对应基准圆点;
根据所述基准圆点、 坐标系、 以及所述标定板的参数完成对所有标志 点的排序, 得到一一对应的图像坐标与世界坐标; 根据所述图像坐标与世界坐标, 完成摄像机参数的标定计算。
[权利要求 5] 根据权利要求 4所述的摄像机标定方法, 其特征在于, 在所述对所述 摄像机采集的标定板图像进行边缘提取之前还包括:
所述摄像机获取标定板图像, 对所述标定板图像进行图像质量预处理
[权利要求 6] 根据权利要求 4所述的摄像机标定方法, 其特征在于, 所述对所述摄 像机采集的标定板图像进行边缘提取包括:
对所述标定板图像进行边缘检测, 得到边缘轮廓的点集合, 筛选剔除 掉所述边缘轮廓中的异常边缘。
[权利要求 7] 根据权利要求 4所述的摄像机标定方法, 其特征在于, 所述对所述摄 像机采集的标定板图像进行边缘提取包括:
首先对所述标定板图像进行像素级的边缘提取, 再对得到的边缘进行 亚像素级定位。
[权利要求 8] 根据权利要求 4所述的摄像机标定方法, 其特征在于, 所述对每个圆 点的边缘轮廓进行椭圆拟合包括:
使用鲁棒的椭圆拟合算法对每个圆点的边缘轮廓进行椭圆拟合。
[权利要求 9] 根据权利要求 4所述的摄像机标定方法, 其特征在于, 所述根据所述 基准圆点、 坐标系、 以及所述标定板的参数完成对所有标志点的排序 , 得到一一对应的图像坐标与世界坐标包括:
以基准标志点作为种子点, 采用基于生长的标志点排序算法, 确定基 准坐标系下所有标志点的方位, 结合所述标定板的参数得到 -对应 的图像坐标与世界坐标。
[权利要求 10] 根据权利要求 4所述的摄像机标定方法, 其特征在于, 所述根据所述 图像坐标与世界坐标, 完成摄像机参数的标定计算包括:
根据所述图像坐标与世界坐标, 计算出理想状态下无畸变的摄像机参 数, 作为摄像机参数初值;
通过最优化方法最小化重投影误差, 迭代求出精确的摄像机参数, 完 成对所述摄像机的参数标定。
[权利要求 11] 一种摄像机, 其特征在于, 所述摄像机包括处理器, 所述处理器用于 执行存储器中存储的计算机程序时实现如权利要求 4至 10任一项所述 的摄像机标定方法。
PCT/CN2018/120963 2018-12-13 2018-12-13 摄像机标定板、标定方法以及摄像机 WO2020118637A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2018/120963 WO2020118637A1 (zh) 2018-12-13 2018-12-13 摄像机标定板、标定方法以及摄像机

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2018/120963 WO2020118637A1 (zh) 2018-12-13 2018-12-13 摄像机标定板、标定方法以及摄像机

Publications (1)

Publication Number Publication Date
WO2020118637A1 true WO2020118637A1 (zh) 2020-06-18

Family

ID=71075890

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/120963 WO2020118637A1 (zh) 2018-12-13 2018-12-13 摄像机标定板、标定方法以及摄像机

Country Status (1)

Country Link
WO (1) WO2020118637A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112985265A (zh) * 2021-04-20 2021-06-18 苏州维嘉科技股份有限公司 线阵相机及其精度补偿方法、存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050207640A1 (en) * 2001-04-02 2005-09-22 Korea Advanced Institute Of Science And Technology Camera calibration system using planar concentric circles and method thereof
CN1801896A (zh) * 2006-01-17 2006-07-12 东南大学 摄像机标定数据的采集方法及其标定板
CN101686406A (zh) * 2008-09-28 2010-03-31 新奥特(北京)视频技术有限公司 一种标定参数的获取方法和装置
CN105264877A (zh) * 2013-04-08 2016-01-20 全视技术有限公司 用于360度摄像机系统的校准的系统和方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050207640A1 (en) * 2001-04-02 2005-09-22 Korea Advanced Institute Of Science And Technology Camera calibration system using planar concentric circles and method thereof
CN1801896A (zh) * 2006-01-17 2006-07-12 东南大学 摄像机标定数据的采集方法及其标定板
CN101686406A (zh) * 2008-09-28 2010-03-31 新奥特(北京)视频技术有限公司 一种标定参数的获取方法和装置
CN105264877A (zh) * 2013-04-08 2016-01-20 全视技术有限公司 用于360度摄像机系统的校准的系统和方法

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112985265A (zh) * 2021-04-20 2021-06-18 苏州维嘉科技股份有限公司 线阵相机及其精度补偿方法、存储介质
CN112985265B (zh) * 2021-04-20 2021-07-30 苏州维嘉科技股份有限公司 线阵相机及其精度补偿方法、存储介质

Similar Documents

Publication Publication Date Title
CN111260731B (zh) 一种棋盘格亚像素级角点自适应检测的方法
CN109829948A (zh) 摄像机标定板、标定方法以及摄像机
WO2017092631A1 (zh) 鱼眼图像的畸变图像校正方法及鱼眼相机的标定方法
WO2020010945A1 (zh) 图像处理方法和装置、电子设备、计算机可读存储介质
CN112223285B (zh) 一种基于组合测量的机器人手眼标定方法
CN109285194B (zh) 相机标定板及相机标定数据采集方法
CN111292379A (zh) 深度相机校正装置及其方法
CN107356213B (zh) 滤光片同心度测量方法及终端设备
CN108716890A (zh) 一种基于机器视觉的高精度尺寸检测方法
CN112132907A (zh) 一种相机标定方法、装置、电子设备及存储介质
CN109636849B (zh) 一种工件定位方法、装置、计算机及计算机可读存储介质
CN112611343A (zh) 一种用于5g环形器的芯片平整度检测装置及其检测方法
JP2005003463A (ja) キャリブレーションチャート画像表示装置、キャリブレーション装置、キャリブレーション方法
CN114463442A (zh) 一种非同轴相机的标定方法
WO2020118637A1 (zh) 摄像机标定板、标定方法以及摄像机
WO2020014845A1 (zh) 一种标定板、内参数标定方法、机器视觉系统及存储装置
JP5486403B2 (ja) 画像処理装置、画像処理方法及びコンピュータプログラム
JP2005521137A (ja) 指紋画像を向上させるための方法
CN110610163A (zh) 一种自然场景下基于椭圆拟合的表格提取方法及工具
TWI747257B (zh) 一種檢查鏤空晶圓的方法
CN114998417A (zh) 基于二次曲线不变量的薄壁冲压件孔组尺寸测量方法
CN112923852B (zh) 基于动态角点定位的sd卡位姿检测方法
CN112383670B (zh) 一种试卷扫描自动扶正方法及装置
CN113723406A (zh) 一种对冠脉造影图像进行支架定位的处理方法和装置
CN110035279B (zh) 在棋盘格测试图中寻找sfr测试区域的方法及装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18942768

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18942768

Country of ref document: EP

Kind code of ref document: A1