WO2018209968A1 - 摄像机标定方法及系统 - Google Patents

摄像机标定方法及系统 Download PDF

Info

Publication number
WO2018209968A1
WO2018209968A1 PCT/CN2017/120321 CN2017120321W WO2018209968A1 WO 2018209968 A1 WO2018209968 A1 WO 2018209968A1 CN 2017120321 W CN2017120321 W CN 2017120321W WO 2018209968 A1 WO2018209968 A1 WO 2018209968A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
module
camera
calibration
icon
Prior art date
Application number
PCT/CN2017/120321
Other languages
English (en)
French (fr)
Inventor
徐一丹
周剑
Original Assignee
成都通甲优博科技有限责任公司
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 成都通甲优博科技有限责任公司 filed Critical 成都通甲优博科技有限责任公司
Publication of WO2018209968A1 publication Critical patent/WO2018209968A1/zh

Links

Images

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 computer vision technology, and in particular to a camera calibration method and system.
  • Camera calibration is a key part of applications such as photogrammetry and machine vision.
  • Image quality is one of the important factors that determine the accuracy and accuracy of post-processing and final results. Distortions are inevitable due to factors such as the lens manufacturing process and the quality of raw materials. These distortions will cause problems such as loss of precision and image distortion in post-processing.
  • the main function of camera calibration is to calculate and estimate the camera's internal parameters and distortion parameters, so that the restored image can be estimated based on these parameters in the post-image processing process to obtain a good quality image.
  • many methods have been proposed for camera calibration.
  • camera calibration methods are generally divided into two categories, namely camera self-calibration method and traditional camera calibration method.
  • the camera self-calibration method does not require a specific calibration reference object, and the camera is calibrated by recording the correspondence between the image and the image of the surrounding environment during the movement of the camera.
  • this kind of calibration methods are: camera self-calibration technology based on active vision (self-calibration technology based on translational motion and self-calibration technology based on rotational motion), camera self-calibration method based on Kruppa equation, hierarchical step-by-step calibration method, based on Self-calibration method for quadric surfaces, etc.
  • the traditional camera calibration is based on a specific camera model, based on a specific calibration reference, through image processing and a series of mathematical transformation methods to obtain the parameters of the camera model.
  • such mature methods include: camera calibration based on 3D stereo calibration (calibration method of camera perspective transformation matrix), camera calibration based on 2D planar calibration (Zhang Zhengyou calibration method), and camera calibration based on radial constraints (Tsai two-step method) and so on.
  • the self-calibration method is more flexible, but it is difficult to obtain stable calibration results due to too many unknown parameters in the calibration process. Moreover, the existing camera self-calibration method generally cannot calibrate the external parameters of the camera. In general, the self-calibration method is mainly used in applications where accuracy is not high (such as communication, virtual reality, etc.). The traditional calibration method is preferred when the accuracy required by the application is high and the parameters of the camera do not change frequently. Based on the camera calibration method of 3D stereo calibration, the 3D stereo calibration used has high requirements on three-dimensional precision and high production cost.
  • the camera calibration accuracy is greatly affected; the camera calibration method based on 2D planar calibration has better robustness and does not require expensive refined 3D calibration, practical Strong, but this method requires the camera to shoot a plane calibration in more than two different orientations, the operation is more complicated.
  • the camera calibration method based on radial constraint has high precision and is suitable for precision measurement. However, this method can not be calibrated to obtain some internal parameters, and the requirements for equipment are also high, which is not suitable for simple calibration.
  • the technical problem to be solved by the present invention is to propose a camera calibration method and system, which solves the problems of complicated operation, low calibration precision and poor robustness of the camera calibration method in the conventional technology.
  • a camera calibration method includes the following steps:
  • the method further includes: capturing a single-board single-icon image of the single-plate;
  • the single-board single-icon setting module only includes one calibration plate; before shooting the single-board single-icon setting module, adjust the relative position of the camera and the single-plate single-setting module: adjust the shooting direction and shooting angle of the camera to be single
  • the board icon is positioned directly in front of the module, and the center of the camera is on the same level as the center of the single image of the board.
  • the image taken by the left camera is called a single-board icon.
  • the left image of the module, the image taken by the right camera is called the single image of the single image of the single image;
  • the multi-board single icon setting module comprises a plurality of calibration plates, wherein the planes of the plurality of calibration plates intersect at two or two; before the multi-board single-icon fixed-module image is captured, adjusting the camera and the multi-board single-icon setting module Relative position: adjust the shooting direction and shooting angle of the camera to the front of the multi-board single icon setting module.
  • the center of the camera is on the same level as the center of the multi-board single-icon setting module. After shooting, the image captured by the left camera is called the multi-board single icon fixed module left image, and the image taken by the right camera is called the multi-board single icon fixed module right image.
  • step b the segmenting the multi-board single icon grouping image specifically includes:
  • the left image and the right image of the multi-board single icon module are respectively divided into N image blocks, wherein each block can be completely contained and only one calibration plate is included, and the image blocks belonging to the same image are sequentially stored in the same A block storage stack.
  • the method further includes: performing feature point detection on the single-plate single-module module image; and performing each block of the multi-board single icon fixed module image and a single-board single-icon setting module
  • the specific method for image feature point detection is to use the Harris algorithm to detect the corner points in the checkerboard as the image feature points, and to find the sub-pixel precision of the feature points, including:
  • I x and I y are the directional derivatives of the horizontal and vertical directions, respectively;
  • the feature points are sequentially numbered in the order of rows or columns.
  • the method for restoring the feature point coordinates to the original calibration module image coordinates includes:
  • the point (o x , o y ) is the coordinate of the origin of the block image in the original calibration module image.
  • step e the calibration of the internal and external parameters of the camera includes:
  • (x w , y w , 0) is the coordinate of the corresponding point of the point (u, v) in the world coordinate in the image
  • r 1 , r 2 , r 3 are respectively three column vectors of the rotation matrix R, s Is a proportional coefficient
  • [x w , y w , 0] T is the coordinate of the point P in the world coordinate system
  • [u, v] T is the ideal image point coordinate of the point P on the image plane
  • R, t are Rotational transformation matrix and translation transformation vector from world coordinate system to camera coordinate system
  • K is the camera internal parameter matrix
  • (u 0 , v 0 ) is the principal point coordinate of the image plane
  • ⁇ , ⁇ are the image in u-axis and v respectively The principal distance parameter of the axis
  • is the image distortion parameter, indicating the skewness of the two coordinate axes of the image
  • is the image distortion parameter, indicating the
  • the coordinates of the points in the image are brought into the equation, and the homography matrix H can be obtained by solving multiple equations in parallel;
  • the equation contains five unknown parameters of the internal parameter matrix, and three homography matrices can generate six equations under the above two constraints, and the images of the three calibration plate planes are brought into the equation to solve the camera.
  • Internal reference matrix K Internal reference matrix
  • the internal parameters and external parameters of the camera can be calculated by steps e1-e4.
  • the camera parameters are optimized, including:
  • the re-projection relationship between the calibration module and the camera image is established. By minimizing the re-projection error, the accurate camera internal reference matrix, distortion matrix, and rotation matrix and translation matrix are obtained.
  • the present invention also provides a camera calibration system, which includes a camera and a calibration module; the calibration module includes a multi-board single icon setting module; and the multi-board single icon fixing module includes a plurality of calibration plates.
  • the planes of the plurality of calibration plates are not all in the same plane;
  • the camera includes an image acquisition module, an image splitting module, a feature point extraction module, a feature point restoration module, and a camera calibration module;
  • the image acquisition module is configured to capture a multi-board single icon fixed module, and obtain a multi-board single icon fixed module image;
  • the image splitting module is configured to block the multi-board single icon fixed module image
  • the feature point extraction module is configured to split each block of the multi-board single icon fixed module image
  • the feature point restoration module is configured to restore each image block feature point coordinate to the original calibration module image coordinate
  • the camera calibration module is configured to calibrate the internal and external parameters of the camera according to the initial parameters of the camera and the initial relative posture information of the camera, in combination with the detected feature point information.
  • the planes of the plurality of calibration plates intersect at two and two.
  • the calibration module further includes a single-board single-icon fixed module; the single-board single-icon fixed module includes only one calibration plate; and the image acquisition module is further configured to capture a single-board single-icon fixed module. Obtaining a single-board single-set module image; the feature point extraction module is further configured to perform feature point detection on the single-board single-icon fixed-module image.
  • FIG. 1 is a schematic diagram of a camera calibration system based on a multi-board single image in Embodiment 1 of the present invention
  • FIG. 2 is a flowchart of a camera calibration method based on a multi-board single icon fixed module according to Embodiment 1 of the present invention
  • FIG. 3 is a flowchart of a camera calibration method based on a hybrid calibration module according to Embodiment 2 of the present invention
  • Figure 4 is a schematic diagram of image segmentation.
  • the invention aims to propose a camera calibration system and method, which solves the problems of complicated operation, low calibration precision and poor robustness of the camera calibration method in the conventional technology.
  • the multi-board single icon fixed module or the multi-board single icon fixed module and the single-board single-icon fixed module are combined to calibrate the camera parameters, thereby reducing a series of errors caused by distortion, and more
  • the board single icon fixing module is composed of a plurality of calibration plates, and the planes of the calibration plates are not all on one plane.
  • the present invention uses an optimized Zhang Zhengyou calibration method to calibrate the internal and external parameters of the camera.
  • FIG. 1 The camera calibration system based on the multi-board single image in this embodiment is shown in FIG. 1 , which includes a camera 1 and a calibration module; the calibration module is a multi-board single-icon fixed module; a calibration plate 2 on the same plane;
  • the camera 1 includes an image acquisition module, an image splitting module, a feature point extraction module, a feature point restoration module, and a camera calibration module;
  • the image acquisition module is configured to capture a multi-board single icon fixed module, and obtain a multi-board single icon fixed module image;
  • the image splitting module is configured to block the multi-board single icon fixed module image
  • the feature point extraction module is configured to perform feature point detection on each block of the multi-board single icon fixed module image splitting;
  • the feature point restoration module is configured to restore each image block feature point coordinate to the original calibration module image coordinate
  • the camera calibration module is configured to calibrate the internal and external parameters of the camera according to the initial parameters of the camera and the initial relative posture information of the camera, in combination with the detected feature point information.
  • the calibration method provided by the embodiment of the present invention is as shown in FIG. 2, and includes the following steps:
  • Step 1 Installing the calibration module:
  • the present invention proposes to use a plurality of calibration plates to form a calibration module.
  • the planes of the calibration plates are not all in the same plane.
  • the planes of the calibration plates intersect at two and two.
  • "Multiple" as used in the present invention means two or more.
  • Step 2 Shooting the calibration module to obtain the calibration module image: using the camera to capture the calibration module, preferably, the shooting direction and angle are directly in front of the module, and the camera center and the module center are on a horizontal surface, which can make the imaging clear .
  • the left camera obtains the image as the left image
  • the right camera obtains the image as the right image (the same below).
  • Step 3 Split the image: divide the image into blocks, and store the image into blocks in a block stack: in the specific implementation, split the left image and the right image into N blocks, wherein each block can Completely included and contains only one calibration plate, blocks belonging to the same image are stored in the same block storage stack in order.
  • the left image is stored in the block stack S1
  • S2 ⁇ A r , B r , C r , D r ⁇ .
  • the blocks belonging to the same image are stored in the same block storage stack in the order in which they are split, which can reduce the time for finding the corresponding block and facilitate the matching between the left image and the right image.
  • Figure 4 illustrates the specific means of splitting the image into blocks: if the calibration module consists of four calibration plates A, B, C, and D, the image obtained by shooting is divided into four blocks, thereby ensuring each image block. Can contain and contain only one calibration plate.
  • Step 4 Feature point detection: Each piece of the image after the split image is subjected to feature point detection, and the feature points are sequentially numbered.
  • the invention detects the corner points in the checkerboard as image feature points and finds the sub-pixel level precision of the feature points. Use the Harris algorithm to detect feature points, as follows:
  • Step 4.1 Using a horizontal and vertical difference operator to filter each pixel of the image to obtain I x and I y , and then obtain the values of four elements in the pixel point n, as follows:
  • I x and I y are the directional derivatives of the horizontal and vertical directions, respectively.
  • Step 4.2 Smooth filtering four elements in pixel n to obtain a new m value.
  • the four elements of n are smoothed using a discrete two-dimensional zero-mean Gaussian filter.
  • the discrete two-dimensional zero-mean Gaussian function is:
  • Step 4.3 Subpixel accuracy is obtained for each feature point q.
  • the specific operation mode is: assume that the point q is near the actual sub-pixel level corner and at the edge of a region A, and the point p is inside the region A, p
  • the gradient is 0, then, by finding many sets of gradients around p points and many vectors These gradients and corresponding vectors
  • the dot product is 0, and then by solving the equations, the solution of the system of equations is the position of the sub-pixel precision of the corner point q, that is, the exact position of the corner point, which is denoted as (x, y).
  • Step 4.4 The feature points are sequentially numbered in the order of rows or columns.
  • Step 5 Restore feature point coordinates: restore the feature point coordinates to the original image coordinates.
  • the coordinates of the corners of the segmented image are the coordinates detected by the computer.
  • the camera calibration is performed on the original entire image. Therefore, the characteristics of the detected image block are required.
  • the point coordinates are restored to their coordinates in the original multi-plate calibration module image. Specifically:
  • the point (o x , o y ) is the coordinate of the origin of the block image in the original image.
  • Step 6 Calibrate the internal and external parameters of the camera and optimize them
  • the internal parameters of the camera describe the parameters that reflect the imaging characteristics of the camera, including the internal reference matrix K, the distortion matrix D, and the camera external parameters are the rotational and translational relationships between the two cameras.
  • the present invention utilizes an improved Zhang Zhengyou calibration method to calibrate the camera. details as follows:
  • Step 6.1 Establish an ideal pinhole imaging model:
  • (x w , y w , 0) is the coordinate of the corresponding point of the point (u, v) in the world coordinates in the image
  • r 1 , r 2 , r 3 are the three column vectors of the rotation matrix R, respectively.
  • Step 6.2 Determine the homography matrix between the two planes: the relationship between the two planes is derived from the pinhole model, that is, the homography matrix is obtained.
  • the homography matrix H can be obtained by taking the coordinates of the points in the image into this equation and solving them by multiple equations (minimum four points).
  • Step 6.3 Solving the internal parameter matrix K by using constraints:
  • this equation contains five unknown parameters of the internal parameter matrix, and three homography matrices can generate six equations under the above two constraints.
  • the camera internal reference matrix K can be solved.
  • Step 6.4 Estimating the external parameters R, T based on the internal parameters:
  • the camera parameters are optimized according to the initial parameters of the camera and the initial relative attitude information of the camera, combined with the detected feature point information.
  • the camera's initial parameters include camera physical focus, cell size, baseline length, and more. Since the camera parameters calibrated by the general method are not ideal, the initial parameters need to be optimized to obtain an accurate camera internal reference matrix, a distortion matrix, and a rotation matrix and a translation matrix.
  • the optimization process is described as: establishing a re-projection relationship between the calibration module and the camera image, by minimizing the re-projection error, and finally obtaining an accurate camera internal reference matrix and distortion matrix, and a rotation matrix and a translation matrix.
  • the camera calibration system provided in this embodiment is a hybrid calibration system based on a multi-board single image and a single-board single image, which includes a camera and a calibration module.
  • the calibration module is the multi-board described in Embodiment 1.
  • a single-board single-icon setting module is also added, and the single-board single-image module only includes one calibration plate;
  • the camera includes an image acquisition module, an image splitting module, a feature point extraction module, a feature point restoration module, and a camera calibration module;
  • the image acquisition module is configured to capture a multi-board single icon fixed module, obtain a multi-board single icon fixed module image, and capture a single-board single-icon fixed module, and obtain a single-board single-icon fixed-module image;
  • the image splitting module is configured to block the multi-board single icon fixed module image
  • the feature point extraction module is configured to perform feature point extraction on each block of the multi-board single-icon fixed-module image splitting and the single-board single-icon fixed-module image;
  • the feature point restoration module is configured to restore each image block feature point coordinate to the original calibration module image coordinate
  • the camera calibration module is configured to calibrate the internal and external parameters of the camera according to the initial parameters of the camera and the initial relative posture information of the camera, in combination with the detected feature point information.
  • the hybrid calibration method provided by the embodiment of the present invention is as shown in FIG. 3 .
  • the module image of the single-board single-picture module and the single-board single-picture module are obtained.
  • Step 1 Shoot the multi-board single icon fixed module image:
  • Step 2 Capture the image of the single-board module:
  • Step 3 split the multi-board single icon to set the module image
  • Step 4 Perform feature point detection on each image block of the split multi-board single image and the single-plate single-picture module image:
  • Step 5 Restore the feature point coordinates of each block to the original image coordinates
  • Step 6 Calibrate the internal and external parameters of the camera and optimize them.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Studio Devices (AREA)
  • Image Processing (AREA)

Abstract

一种摄像机(1)标定方法及系统,解决了传统技术中的摄像机(1)标定方法存在的操作繁琐复杂、标定精度低、鲁棒性差的问题。该方法包括以下步骤:a.安装由多个标定板(2)组成的标定模组;b.拍摄标定模组,获取标定模组图像;c.对标定组图像进行分块;d.对每一个图像块进行特征点检测;e.将特征点坐标还原到原标定模组图像坐标;f.对相机参数进行优化。

Description

摄像机标定方法及系统
本申请要求于2017年05月16日提交的申请号为201710344476.0、名称为“一种摄像机标定方法及系统”的中国专利申请的优先权,并将其全部内容通过引用的方式结合在本申请中。
技术领域
本发明涉及计算机视觉技术,具体涉及摄像机标定方法及系统。
背景技术
摄像机标定是摄影测量以及机器视觉等应用中的一个关键环节,影像的质量是决定后期数据处理与最终成果精度与准确性的重要因素之一。由于镜头制作工艺、原材料质量等因素的限制,不可避免地出现畸变,这些畸变将导致后期处理中产生诸如精度损失、影像形变等问题。
相机标定的主要作用是计算以及估计相机的内部参数,畸变参数,使得在后期图像处理过程中能够根据这些参数估计还原图像,得到质量好的图像。迄今为止,对于摄像机标定问题已经提出了很多方法,一般的,摄像机标定方法一般分为两类,即摄像机自标定方法和传统摄像机标定方法。摄像机自标定方法不需要特定的标定参照物,通过记录摄像机在运动过程中周围环境的图像与图像之间的对应关系来对摄像机进行标定。目前这一类标定方法有:基于主动视觉的摄像机自标定技术(基于平移运动的自标定技术和基于旋转运动的自标定技术),基于Kruppa方程的摄像机自标定方法,分层逐步标定法,基于二次曲面的自标定方法等。传统的摄像机标定是在一定的摄像机模型下,基于特定的标定参照物,通过对其进行图像处理以及利用一系列数学变换方法,求取摄像机模型的参数。目前这类成熟的表定方法包括:基于3D立体标定物的摄像机标定(摄像机透视变换矩阵的标定方法)、基于2D平面标定物的摄像机标定(张正友标定法),以及基于径向约束的摄像机标定(Tsai两步法)等。
自标定方法比较灵活,但是由于标定过程中未知参数过多,所以很难得到稳定的标定结果。并且,已有的摄像机自标定方法一般无法标定出摄像机外部参数。一般来说,自标定方法主要应用于精度要求不高的场合(如通讯、虚拟现实等)。 当应用场合所要求的精度很高且摄像机的参数不经常变化时,传统标定方法为首选。基于3D立体标定物的摄像机标定方法,所采用的3D立体标定物对于三维精度要求很高,制作成本较高。由于在实施过程中忽略了摄像机的非线性畸变,导致摄像机标定精度受到很大影响;基于2D平面标定物的摄像机标定方法具有较好的鲁棒性,并且不需昂贵的精制3D标定物,实用性较强,但是该方法要求摄像机在两个以上不同的方位拍摄一个平面标定物,操作较复杂。基于径向约束的摄像机标定方法的精度比较高,适用于精密测量,但此方法不能具体标定得出部分内部参数,并且对设备的要求也很高,不适用于简单的标定。
发明内容
本发明所要解决的技术问题是:提出一种摄像机标定方法及系统,解决传统技术中的摄像机标定方法存在的操作繁琐复杂、标定精度低、鲁棒性差的问题。
本发明解决上述技术问题所采用的方案是:
一种摄像机标定方法,包括以下步骤:
a.拍摄多板单图标定模组图像;
b.对多板单图标定模组图像进行分块;
c.对多板单图标定模组图像拆分的每一个块进行特征点检测;
d.将特征点坐标还原到原标定模组图像坐标;
e.标定相机内外参数并进行优化。
作为进一步优化,步骤a中,还包括:拍摄单板单图标定模组图像;
所述单板单图标定模组仅包含一个标定板;在拍摄单板单图标定模组前,调整摄像机与单板单图标定模组的相对位置:调整摄像机的拍摄方向和拍摄角度为单板单图标定模组正前方,摄像机的中心与单板单图的中心在同一水平面上;在对单板单图标定模组进行拍摄后,通过左相机拍摄的图像称为单板单图标定模组左图像,通过右相机拍摄的图像称为单板单图标定模组右图像;
所述多板单图标定模组包含多个标定板,所述多个标定板所在平面两两相交;在拍摄多板单图标定模组图像前,调整摄像机与多板单图标定模组的相对位置:调整摄像机的拍摄方向和拍摄角度为多板单图标定模组正前方,摄像机的中心与多板单图标定模组的中心在同一水平面上;在对多板单图标定模组进行拍摄后, 通过左相机拍摄的图像称为多板单图标定模组左图像,通过右相机拍摄的图像称为多板单图标定模组右图像。
作为进一步优化,步骤b中,所述对多板单图标定组图像进行分块具体包括:
将多板单图标定模组左图像和右图像分别拆分成N个图像块,其中,每个块都能够完整包含且仅仅包含一个标定板,属于同一张图像的图像块按顺序存入同一个块存储栈。
作为进一步优化,步骤c中,还包括:对单板单图标定模组图像进行特征点检测;对所述多板单图标定模组图像拆分的每一个块及单板单图标定模组图像进行特征点检测的具体方法是利用Harris算法通过检测棋盘格内角点作为图像特征点,并求特征点亚像素级精度,具体包括:
c1.利用水平、竖直差分算子对图像每个像素进行滤波以求得I x、I y,进而求得像素点n中四个元素的值,如下:
Figure PCTCN2017120321-appb-000001
其中,
Figure PCTCN2017120321-appb-000002
其中,I x、I y分别是水平和垂直方向的方向导数;
c2.利用离散二维零均值高斯滤波器对像素点n中的四个元素进行平滑滤波,离散二维零均值高斯函数为:
Figure PCTCN2017120321-appb-000003
c3.对每一个特征点q求亚像素级精度:假设点q在实际亚像素级角点的附近,且在一个区域A的边缘,点p在区域A的内部,p处的梯度为0,那么,通过找到p点周围多组梯度以及多个向量
Figure PCTCN2017120321-appb-000004
这些梯度与对应的向量
Figure PCTCN2017120321-appb-000005
的点积为0,然后通过求解方程组,方程组的解即为角点q的亚像素级精度的位置,也就是角点的精确位置,记为(x,y);
c4.将特征点按行或列的顺序进行顺序编号。
作为进一步优化,步骤d中,所述将特征点坐标还原到原标定模组图像坐标的方法包括:
对于每一块图像上检测到的特征点p(x,y),通过以下关系找到其在原标定模组图像的坐标P(X,Y):
Figure PCTCN2017120321-appb-000006
其中,点(o x,o y)为分块图像的原点在原标定模组图像中的坐标。
作为进一步优化,步骤e中,所述标定相机内外参数,具体包括:
e1.建立理想针孔成像模型:
Figure PCTCN2017120321-appb-000007
其中,(x w,y w,0)为图像中点(u,v)在世界坐标中的对应点的坐标,r 1,r 2,r 3分别为旋转矩阵R的三个列向量,s是一个比例系数,[x w,y w,0] T为点P在世界坐标系下的坐标;[u,v] T为点P在图像平面上的理想像点坐标;R、t分别为从世界坐标系到相机坐标系的旋转变换矩阵和平移变换向量;K为相机内部参数矩阵;(u 0,v 0)为图像平面的主点坐标;α,β分别为图像在u轴和v轴的主距参数;γ为图像扭曲参数,表示图像两坐标轴偏斜度;
e2.确定两个平面之间的单应矩阵:由针孔成像模型推算出两个平面之间的关系,即求单应矩阵:
H=[h 1 h 2 h 3]=K[r 1 r 2 0],则两个平面内的对应点的对应关系为:
Figure PCTCN2017120321-appb-000008
将图像中的点的坐标带入此方程,由多个方程联立求解即可求出单应矩阵H;
e3.利用约束条件求解内参矩阵K:
由[h 1 h 2 h 3]=λK[r 1 r 2 t],以及旋转列向量r 1,r 2的单位正交,旋转向量 模为1,得到如下约束方程:
Figure PCTCN2017120321-appb-000009
其中,该方程包含内参矩阵的5个未知参数,由3个单应矩阵在上述2个约束条件下可产生6个方程,将三个标定板平面的图像带入此方程,即可解出相机内参矩阵K;
e4.基于内参矩阵估算外部参数R,T:
相机外部参数表示相机内摄像头之间的旋转平移关系,由[h 1 h 2 h 3]=λK[r 1 r 2 t],可以解出:
Figure PCTCN2017120321-appb-000010
通过步骤e1-e4即可计算获得相机的内部参数和外部参数。
作为进一步优化,步骤e中,所述对相机参数进行优化,具体包括:
建立标定模组与相机图像之间的重投影关系,通过最小化重投影误差,最后得到精准的相机内参矩阵、畸变矩阵、以及旋转矩阵和平移矩阵。
此外,本发明还提出了一种摄像机标定系统,其包括摄像机、标定模组;所述标定模组包括多板单图标定模组;所述多板单图标定模组包括多个标定板,所述多个标定板所在平面不都在同一个平面;
所述摄像机包括图像获取模块、图像拆分模块、特征点提取模块、特征点还原模块、相机标定模块;
所述图像获取模块,用于拍摄多板单图标定模组,获取多板单图标定模组图像;
所述图像拆分模块,用于对多板单图标定模组图像进行分块;
所述特征点提取模块,用于对多板单图标定模组图像拆分的每一个块;
所述特征点还原模块,用于将每一个图像块特征点坐标还原到原标定模组图像坐标;
相机标定模块,用于根据相机的初始参数以及相机的初始相对姿态信息,结合检测到的特征点信息,标定相机内外参数。
作为进一步优化,所述多个标定板所在平面两两相交。
作为进一步优化,所述标定模组还包括单板单图标定模组;所述单板单图标定模组仅含一个标定板;所述图像获取模块还用于拍摄单板单图标定模组,获取单板单图标定模组图像;所述特征点提取模块还用于对单板单图标定模组图像进行特征点检测。
本发明的有益效果是:
1)利用若干标定板组成一个多板单图标定模组,使得标定过程简单,标定精度高;
2)多板单图标定模组内的多个标定板所在平面不都在一个平面上,解决了传统方法中需要多次多角度拍摄标定板从而完成标定的麻烦问题,同时,降低了由于多次操作引起的人为操作误差的可能性;
3)利用多板单图标定模组与单板单图标定模组相结合的方法对相机参数进行标定,降低了由于畸变引起的一系列误差。
附图说明
图1为本发明实施例1中的基于多板单图的摄像机标定系统示意图;
图2为本发明实施例1中的基于多板单图标定模组的摄像机标定方法流程图;
图3为本发明实施例2中的基于混合标定模组的摄像机标定方法流程图;
图4为图像分块示意图。
具体实施方式
本发明旨在提出一种摄像机标定系统及方法,解决传统技术中的摄像机标定方法存在的操作繁琐复杂、标定精度低、鲁棒性差的问题。在本发明中,采用多板单图标定模组或者多板单图标定模组与单板单图标定模组相结合的方法对相机参数进行标定,降低了由于畸变引起的一系列误差,多板单图标定模组由多个标定板构成,且标定板所在的平面不都在一个平面上,在拍摄标定模组时,只需拍摄一次,从而避免传统技术中多次多角度拍摄标定板的繁琐步骤,也降低了人为操作误差的可能性,在获取多板单图标定模组图像后对图像进行拆分,然后对拆分后的图像块及单板单图模组图像进行特征提取,并将特征点坐标还原到原标 定模组图像坐标,最后,本发明采用一种优化的张正友标定法标定相机内外参数。
下面结合附图及实施例对本发明的方案作进一步的描述:
实施例1:
本实施例中的基于多板单图的摄像机标定系统如图1所示,其包括摄像机1、标定模组;所述标定模组为多板单图标定模组;包括4个所在平面不都在同一个平面上的标定板2;
所述摄像机1包括图像获取模块、图像拆分模块、特征点提取模块、特征点还原模块、相机标定模块;
所述图像获取模块,用于拍摄多板单图标定模组,获取多板单图标定模组图像;
所述图像拆分模块,用于对多板单图标定模组图像进行分块;
所述特征点提取模块,用于对多板单图标定模组图像拆分的每一个块进行特征点检测;
所述特征点还原模块,用于将每一个图像块特征点坐标还原到原标定模组图像坐标;
相机标定模块,用于根据相机的初始参数以及相机的初始相对姿态信息,结合检测到的特征点信息,标定相机内外参数。
基于上述摄像机标定系统,本发明实施例提供的标定方法如图2所示,包括以下步骤:
步骤1、安装标定模组:为了解决传统标定方法的标定过程中因需要多个角度移动拍摄标定板引起的标定误差和操作复杂度,本发明提出利用多个标定板组成一个标定模组,这些标定板所在平面不都在同一个平面,优选的,这些标定板所在平面两两相交。本发明中所述“多个”是指两个或两个以上。
步骤2、拍摄标定模组,获取标定模组图像:利用摄像机拍摄标定模组,优选的,拍摄方向和角度为模组正前方,摄像机中心与模组中心在一个水平面上,这样能够使得成像清晰。其中,左相机获得图像称为左图像,右相机获得图像称为右图像(下同)。
步骤3、拆分图像:对图像进行分块,并将图像分块按顺序存入一个块栈中:具体实现时,将左图像和右图像分别拆分成N块,其中,每块都能够完整包含 且仅仅包含一个标定板,属于同一张图像的块按顺序存入同一个块存储栈中。本实施例将左图像分块存入块栈S1,右图像分块存入块栈S2,S1={A l,B l,C l,D l},S2={A r,B r,C r,D r}。对属于同一张图像的块按拆分出来的先后顺序存入同一个块存储栈中,能够能够减少寻找对应块的时间,方便左图像和右图像之间的匹配。
图4示意了对图像分块拆分的具体手段:若标定模组由A、B、C、D四个标定板组成,则对应将拍摄获取的图像分成4块,由此保证每个图像块能够包含且仅仅包含一个标定板。
步骤4、特征点检测:将拆分后的图像的每一块图像进行特征点检测,并将特征点进行顺序编号。本发明通过检测棋盘格内角点作为图像特征点,并求特征点亚像素级精度。利用Harris算法检测特征点,具体如下:
步骤4.1、利用水平、竖直差分算子对图像每个像素进行滤波以求得I x、I y,进而求得像素点n中四个元素的值,如下:
Figure PCTCN2017120321-appb-000011
其中,
Figure PCTCN2017120321-appb-000012
其中,I x、I y分别是水平和垂直方向的方向导数。
步骤4.2、对像素点n中的四个元素进行平滑滤波,得到新的m值。在此,利用离散二维零均值高斯滤波器对n的四个元素进行平滑。离散二维零均值高斯函数为:
Figure PCTCN2017120321-appb-000013
步骤4.3、对每一个特征点q求亚像素级精度,具体操作方式为:假设点q在实际亚像素级角点的附近且在一个区域A的边缘,点p在区域A的内部,p处的梯度为0,那么,通过找到p点周围很多组梯度以及很多个向量
Figure PCTCN2017120321-appb-000014
这些梯度与对应的向量
Figure PCTCN2017120321-appb-000015
的点积为0,然后通过求解方程组,方程组的解即为角点q的亚像素级精度的位置,也就是角点的精确位置,记为(x,y)。
步骤4.4、将特征点按行或列的顺序进行顺序编号。
步骤5、还原特征点坐标:将特征点坐标还原到原图像坐标。在图像分块检测特征点的过程中,分块图像角点的坐标是由计算机检测出的坐标,然而,相机标定是针对原来的整张图像进行,因此,需要将检测到的图像块的特征点坐标还原到其在原来多板标定模组图像的坐标。具体为:
对于每一块图像上检测到的特征点p(x,y),通过以下关系找到其在原来多板标定模组图像的坐标P(X,Y):
Figure PCTCN2017120321-appb-000016
其中,点(o x,o y)为分块图像的原点在原图像的坐标。
步骤6、标定相机内外参数并进行优化;
摄像机内部参数描述的是能够反映摄像机的成像特性的参数,包括内参矩阵K,畸变矩阵D,相机外部参数便是两个摄像头之间的旋转和平移关系。本发明利用一种改进的张正友标定法对相机进行标定。具体如下:
步骤6.1、建立理想针孔成像模型:
Figure PCTCN2017120321-appb-000017
其中,其中,(x w,y w,0)为图像中点(u,v)在世界坐标中的对应点的坐标,r 1,r 2,r 3分别为旋转矩阵R的三个列向量,s是一个比例系数,通过上述方程计算得到;[x w,y w,0] T为点P在世界坐标系下的坐标;[u,v] T为点P在图像平面上的理想像点坐标;R和t为从世界坐标系到相机坐标系的旋转变换矩阵和平移变换向量;K为相机内部参数矩阵;(u 0,v 0)为图像平面的主点坐标;α,β分别为图像在u轴和v轴的主距参数;γ为图像扭曲参数,表示图像两坐标轴偏斜度,并假设标定板平面在世界坐标系Z=0的平面上。
步骤6.2、确定两个平面之间的单应矩阵:由针孔模型推算出两个平面之间的关系,即求单应矩阵。
单应矩阵为:H=[h 1 h 2 h 3]=K[r 1 r 2 0],则两个平面内的对应点的对应关系为:
Figure PCTCN2017120321-appb-000018
将图像中的点的坐标带入此方程,由多个方程联立求解(最少四个点),即可求出单应矩阵H。
步骤6.3、利用约束条件求解内参矩阵K:
由[h 1 h 2 h 3]=λK[r 1 r 2 t],以及旋转列向量r 1,r 2的单位正交,旋转向量模为1,得到如下约束方程:
Figure PCTCN2017120321-appb-000019
其中,这个方程包含内参矩阵的5个未知参数,由3个单应矩阵在上述2个约束条件下可以产生6个方程。将三个标定板平面的图像带入此方程,即可解出相机内参矩阵K。
步骤6.4、基于内参估算外部参数R,T:
相机外部参数表示相机内摄像头之间的旋转平移关系,由[h 1 h 2 h 3]=λK[r 1 r 2 t],可以解出:
Figure PCTCN2017120321-appb-000020
在对相机内外参数标定后,根据相机的初始参数以及相机的初始相对姿态信息,结合检测到的特征点信息,对相机参数进行优化。
相机的初始参数包括相机物理焦距、像元尺寸、基线长度等。由于一般方法标定出的相机参数不理想,因此需要对初始参数进行优化,得到精确的相机内参矩阵、畸变矩阵以及旋转矩阵和平移矩阵。优化过程描述为:建立标定模组与相机图像之间的重投影关系,通过最小化重投影误差,最后得到精准的相机内参矩 阵和畸变矩阵,以及旋转矩阵和平移矩阵。
实施例2:
本实施例中提供的摄像机标定系统,是一种基于多板单图和单板单图的混合标定系统,其包括摄像机和标定模组,此标定模组在实施例1中所述的多板单图的标定模组的基础上,还增加了单板单图标定模组,所述单板单图模组仅包含一个标定板;
所述摄像机包括图像获取模块、图像拆分模块、特征点提取模块、特征点还原模块、相机标定模块;
所述图像获取模块,用于拍摄多板单图标定模组,获取多板单图标定模组图像;以及拍摄单板单图标定模组,获取单板单图标定模组图像;
所述图像拆分模块,用于对多板单图标定模组图像进行分块;
所述特征点提取模块,用于对多板单图标定模组图像拆分的每一个块以及单板单图标定模组图像进行特征点提取;
所述特征点还原模块,用于将每一个图像块特征点坐标还原到原标定模组图像坐标;
相机标定模块,用于根据相机的初始参数以及相机的初始相对姿态信息,结合检测到的特征点信息,标定相机内外参数。
基于上述摄像机标定系统,本发明实施例提供的混合标定方法如图3所示,相比实施例1中的标定方法,其多了获取单板单图模组图像、对单板单图模组图像进行特征点提取的步骤;
本实施例中的混合标定方法具体包括:
步骤1、拍摄多板单图标定模组图像:
步骤2、拍摄单板单图模组图像:
步骤3、拆分多板单图标定模组图像;
步骤4、对拆分的多板单图每一个图像块以及单板单图模组图像进行特征点检测:
步骤5、将每一块图的特征点坐标还原到原图像坐标;
步骤6、标定相机内外参数并进行优化。

Claims (10)

  1. 一种摄像机标定方法,其特征在于,包括以下步骤:
    a.拍摄多板单图标定模组图像;
    b.对多板单图标定模组图像进行分块;
    c.对多板单图标定模组图像拆分的每一个块进行特征点检测;
    d.将特征点坐标还原到原标定模组图像坐标;
    e.标定相机内外参数并进行优化。
  2. 如权利要求1所述的一种摄像机标定方法,其特征在于,步骤a中,还包括:拍摄单板单图标定模组图像;
    所述单板单图标定模组仅包含一个标定板;在拍摄单板单图标定模组前,调整摄像机与单板单图标定模组的相对位置:调整摄像机的拍摄方向和拍摄角度为单板单图标定模组正前方,摄像机的中心与单板单图的中心在同一水平面上;在对单板单图标定模组进行拍摄后,通过左相机拍摄的图像称为单板单图标定模组左图像,通过右相机拍摄的图像称为单板单图标定模组右图像;
    所述多板单图标定模组包含多个标定板,所述多个标定板所在平面两两相交;在拍摄多板单图标定模组图像前,调整摄像机与多板单图标定模组的相对位置:调整摄像机的拍摄方向和拍摄角度为多板单图标定模组正前方,摄像机的中心与多板单图标定模组的中心在同一水平面上;在对多板单图标定模组进行拍摄后,通过左相机拍摄的图像称为多板单图标定模组左图像,通过右相机拍摄的图像称为多板单图标定模组右图像。
  3. 如权利要求1所述的一种摄像机标定方法,其特征在于,步骤b中,所述对多板单图标定组图像进行分块具体包括:
    将多板单图标定模组左图像和右图像分别拆分成N个图像块,其中,每个块都能够完整包含且仅仅包含一个标定板,属于同一张图像的图像块按顺序存入同一个块存储栈。
  4. 如权利要求2所述的一种摄像机标定方法,其特征在于,步骤c中,还包括:对单板单图标定模组图像进行特征点检测;对所述多板单图标定模组图像拆分的每一个块及单板单图标定模组图像进行特征点检测的具体方法是利用 Harris算法通过检测棋盘格内角点作为图像特征点,并求特征点亚像素级精度,具体包括:
    c1.利用水平、竖直差分算子对图像每个像素进行滤波以求得I x、I y,进而求得像素点n中四个元素的值,如下:
    Figure PCTCN2017120321-appb-100001
    其中,
    Figure PCTCN2017120321-appb-100002
    其中,I x、I y分别是水平和垂直方向的方向导数;
    c2.利用离散二维零均值高斯滤波器对像素点n中的四个元素进行平滑滤波,离散二维零均值高斯函数为:
    Figure PCTCN2017120321-appb-100003
    c3.对每一个特征点q求亚像素级精度:假设点q在实际亚像素级角点的附近,且在一个区域A的边缘,点p在区域A的内部,p处的梯度为0,那么,通过找到p点周围多组梯度以及多个向量
    Figure PCTCN2017120321-appb-100004
    这些梯度与对应的向量
    Figure PCTCN2017120321-appb-100005
    的点积为0,然后通过求解方程组,方程组的解即为角点q的亚像素级精度的位置,也就是角点的精确位置,记为(x,y);
    c4.将特征点按行或列的顺序进行顺序编号。
  5. 如权利要求1所述的一种摄像机标定方法,其特征在于,
    步骤d中,所述将特征点坐标还原到原标定模组图像坐标的方法包括:
    对于每一块图像上检测到的特征点p(x,y),通过以下关系找到其在原标定模组图像的坐标P(X,Y):
    Figure PCTCN2017120321-appb-100006
    其中,点(o x,o y)为分块图像的原点在原标定模组图像中的坐标。
  6. 如权利要求1所述的一种摄像机标定方法,其特征在于,步骤e中,所述标定相机内外参数,具体包括:
    e1.建立理想针孔成像模型:
    Figure PCTCN2017120321-appb-100007
    其中,(x w,y w,0)为图像中点(u,v)在世界坐标中的对应点的坐标,r 1,r 2,r 3分别为旋转矩阵R的三个列向量,s是一个比例系数,[x w,y w,0] T为点P在世界坐标系下的坐标;[u,v] T为点P在图像平面上的理想像点坐标;R、t分别为从世界坐标系到相机坐标系的旋转变换矩阵和平移变换向量;K为相机内部参数矩阵;(u 0,v 0)为图像平面的主点坐标;α,β分别为图像在u轴和v轴的主距参数;γ为图像扭曲参数,表示图像两坐标轴偏斜度;
    e2.确定两个平面之间的单应矩阵:由针孔成像模型推算出两个平面之间的关系,即求单应矩阵:
    H=[h 1 h 2 h 3]=K[r 1 r 2 0],则两个平面内的对应点的对应关系为:
    Figure PCTCN2017120321-appb-100008
    将图像中的点的坐标带入此方程,由多个方程联立求解即可求出单应矩阵H;
    e3.利用约束条件求解内参矩阵K:
    由[h 1 h 2 h 3]=λK[r 1 r 2 t],以及旋转列向量r 1,r 2的单位正交,旋转向量模为1,得到如下约束方程:
    Figure PCTCN2017120321-appb-100009
    其中,该方程包含内参矩阵的5个未知参数,由3个单应矩阵在上述2个约束条件下可产生6个方程,将三个标定板平面的图像带入此方程,即可解出相机内参矩阵K;
    e4.基于内参矩阵估算外部参数R,T:
    相机外部参数表示相机内摄像头之间的旋转平移关系,由[h 1 h 2 h 3]=λK[r 1 r 2 t],可以解出:
    Figure PCTCN2017120321-appb-100010
    通过步骤e1-e4即可计算获得相机的内部参数和外部参数。
  7. 如权利要求6所述的一种摄像机标定方法,其特征在于,步骤e中,所述对相机参数进行优化,具体包括:
    建立标定模组与相机图像之间的重投影关系,通过最小化重投影误差,最后得到精准的相机内参矩阵、畸变矩阵、以及旋转矩阵和平移矩阵。
  8. 一种摄像机标定系统,其包括摄像机、标定模组,其特征在于;所述标定模组包括多板单图标定模组;所述多板单图标定模组包括多个标定板,所述多个标定板所在平面不都在同一个平面;
    所述摄像机包括图像获取模块、图像拆分模块、特征点提取模块、特征点还原模块、相机标定模块;
    所述图像获取模块,用于拍摄多板单图标定模组,获取多板单图标定模组图像;
    所述图像拆分模块,用于对多板单图标定模组图像进行分块;
    所述特征点提取模块,用于对多板单图标定模组图像拆分的每一个块;
    所述特征点还原模块,用于将每一个图像块特征点坐标还原到原标定模组图像坐标;
    相机标定模块,用于根据相机的初始参数以及相机的初始相对姿态信息,结合检测到的特征点信息,标定相机内外参数。
  9. 如权利要求8所述的一种摄像机标定系统,其特征在于,所述多个标定板所在平面两两相交。
  10. 如权利要求8所述的一种摄像机标定系统,其特征在于,所述标定模组还包括单板单图标定模组;所述单板单图标定模组仅含一个标定板;所述图像获取模块还用于拍摄单板单图标定模组,获取单板单图标定模组图像;所述特征点提取模块还用于对单板单图标定模组图像进行特征点检测。
PCT/CN2017/120321 2017-05-16 2017-12-29 摄像机标定方法及系统 WO2018209968A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710344476.0 2017-05-16
CN201710344476.0A CN107194972B (zh) 2017-05-16 2017-05-16 一种摄像机标定方法及系统

Publications (1)

Publication Number Publication Date
WO2018209968A1 true WO2018209968A1 (zh) 2018-11-22

Family

ID=59872649

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/120321 WO2018209968A1 (zh) 2017-05-16 2017-12-29 摄像机标定方法及系统

Country Status (2)

Country Link
CN (1) CN107194972B (zh)
WO (1) WO2018209968A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109636859A (zh) * 2018-12-24 2019-04-16 武汉大音科技有限责任公司 一种基于单相机的三维视觉检测的标定方法

Families Citing this family (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107194972B (zh) * 2017-05-16 2021-04-02 成都通甲优博科技有限责任公司 一种摄像机标定方法及系统
CN108230396A (zh) * 2017-12-08 2018-06-29 深圳市商汤科技有限公司 标定装置和标定设备
CN107993268B (zh) * 2017-12-26 2021-11-30 广东工业大学 一种相机自标定的方法及系统
CN108334199A (zh) * 2018-02-12 2018-07-27 华南理工大学 基于增强现实的移动式多模态交互方法及装置
CN108986170B (zh) * 2018-07-03 2020-11-06 武汉精测电子集团股份有限公司 一种适用于现场工况的线阵相机平场校正方法
JP7219561B2 (ja) * 2018-07-18 2023-02-08 日立Astemo株式会社 車載環境認識装置
CN109389649B (zh) * 2018-09-27 2020-11-03 北京伟景智能科技有限公司 双目相机的快速标定方法和装置
CN110264524B (zh) * 2019-05-24 2023-07-21 联想(上海)信息技术有限公司 一种标定方法、装置、系统及存储介质
CN110189380B (zh) * 2019-05-30 2021-12-07 Oppo广东移动通信有限公司 标定数据的优化方法、结构光模组及存储介质
CN112116662A (zh) * 2019-06-20 2020-12-22 上海炬佑智能科技有限公司 摄像头校正用设备
CN110517321B (zh) * 2019-07-10 2022-05-17 湖南交工智能技术有限公司 相机标定方法、相机及存储介质
CN112785650B (zh) * 2019-11-11 2024-01-12 北京京邦达贸易有限公司 一种相机参数标定方法及装置
CN112907676B (zh) * 2019-11-19 2022-05-10 浙江商汤科技开发有限公司 传感器的标定方法、装置、系统、车辆、设备及存储介质
CN112907675B (zh) * 2019-11-19 2022-05-24 浙江商汤科技开发有限公司 图像采集设备的标定方法、装置、系统、设备及存储介质
TWI720869B (zh) * 2020-04-15 2021-03-01 致伸科技股份有限公司 相機模組之對位方法
CN113538564A (zh) * 2020-04-17 2021-10-22 致伸科技股份有限公司 相机模块的对位方法
CN112611361A (zh) * 2020-12-08 2021-04-06 华南理工大学 一种无人机机载测绘吊舱的摄像头安装误差测定方法
CN112927301B (zh) * 2021-02-04 2024-09-13 深圳市杉川机器人有限公司 摄像头标定方法、装置、计算设备及可读存储介质
CN112991466B (zh) * 2021-04-26 2021-09-10 杭州灵西机器人智能科技有限公司 一种基于不完整标定板的相机标定的方法和系统
CN114419168B (zh) * 2022-01-21 2022-09-09 北京微链道爱科技有限公司 一种图像特征点的标定方法及装置

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090141043A1 (en) * 2007-11-30 2009-06-04 Hitachi, Ltd. Image mosaicing apparatus for mitigating curling effect
CN102135236A (zh) * 2011-01-05 2011-07-27 北京航空航天大学 双目视觉管道内壁自动无损检测方法
CN103413319A (zh) * 2013-08-27 2013-11-27 哈尔滨工业大学 一种工业摄像机参数现场标定方法
CN106548477A (zh) * 2017-01-24 2017-03-29 长沙全度影像科技有限公司 一种基于立体标定靶的多路鱼眼相机标定装置及方法
CN107194972A (zh) * 2017-05-16 2017-09-22 成都通甲优博科技有限责任公司 一种摄像机标定方法及系统

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1851752A (zh) * 2006-03-30 2006-10-25 东南大学 三维重构系统中双摄像机标定方法
CN100557635C (zh) * 2008-06-10 2009-11-04 北京航空航天大学 一种基于柔性立体靶标的摄像机标定方法
CN102065313B (zh) * 2010-11-16 2012-10-31 上海大学 平行式相机阵列的未标定多视点图像校正方法
KR20130035422A (ko) * 2011-09-30 2013-04-09 삼성전기주식회사 카메라 모듈의 보정 장치
US9596459B2 (en) * 2014-09-05 2017-03-14 Intel Corporation Multi-target camera calibration
CN105528785B (zh) * 2015-12-03 2018-06-15 河北工业大学 一种双目视觉图像立体匹配方法
CN106600648A (zh) * 2016-12-06 2017-04-26 合肥工业大学 标定相机内参数和畸变系数的立体编码靶标及其标定方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090141043A1 (en) * 2007-11-30 2009-06-04 Hitachi, Ltd. Image mosaicing apparatus for mitigating curling effect
CN102135236A (zh) * 2011-01-05 2011-07-27 北京航空航天大学 双目视觉管道内壁自动无损检测方法
CN103413319A (zh) * 2013-08-27 2013-11-27 哈尔滨工业大学 一种工业摄像机参数现场标定方法
CN106548477A (zh) * 2017-01-24 2017-03-29 长沙全度影像科技有限公司 一种基于立体标定靶的多路鱼眼相机标定装置及方法
CN107194972A (zh) * 2017-05-16 2017-09-22 成都通甲优博科技有限责任公司 一种摄像机标定方法及系统

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109636859A (zh) * 2018-12-24 2019-04-16 武汉大音科技有限责任公司 一种基于单相机的三维视觉检测的标定方法
CN109636859B (zh) * 2018-12-24 2022-05-10 武汉大音科技有限责任公司 一种基于单相机的三维视觉检测的标定方法

Also Published As

Publication number Publication date
CN107194972A (zh) 2017-09-22
CN107194972B (zh) 2021-04-02

Similar Documents

Publication Publication Date Title
WO2018209968A1 (zh) 摄像机标定方法及系统
CN110689579B (zh) 基于合作目标的快速单目视觉位姿测量方法及测量系统
CN107633536B (zh) 一种基于二维平面模板的相机标定方法及系统
US10559090B2 (en) Method and apparatus for calculating dual-camera relative position, and device
CN108230397B (zh) 多目摄像机标定与校正方法和装置、设备、程序和介质
CN107507235B (zh) 一种基于rgb-d设备采集的彩色图像和深度图像的配准方法
CN109360246B (zh) 基于同步子区搜索的立体视觉三维位移测量方法
CN107886547B (zh) 一种鱼眼相机标定方法及系统
CN108629810B (zh) 双目相机的标定方法、装置及终端
CN108564617A (zh) 多目相机的三维重建方法、装置、vr相机和全景相机
CN103971378A (zh) 一种混合视觉系统中全景图像的三维重建方法
CN113592721B (zh) 摄影测量方法、装置、设备及存储介质
CN108734738B (zh) 相机标定方法及装置
CN107635135B (zh) 双摄模组组装前相对倾斜角度的测试方法及测试系统
CN113450416B (zh) 一种应用于三目相机立体标定的tcsc方法
CN104167001B (zh) 基于正交补偿的大视场摄像机标定方法
CN114998447A (zh) 多目视觉标定方法及系统
CN107067441B (zh) 摄像机标定方法及装置
CN114998773A (zh) 适用于无人机系统航拍图像的特征误匹配剔除方法及系统
CN112164119B (zh) 一种适用于狭小空间的环绕放置多摄像机系统标定方法
RU2692970C2 (ru) Способ калибровки видеодатчиков многоспектральной системы технического зрения
CN111968182A (zh) 一种双目相机非线性模型参数的标定方法
JP7033294B2 (ja) 撮像システム、撮像方法
CN111091595B (zh) 斜视三维测绘方法及测绘系统
CN103310448B (zh) 用于das的摄像装置姿态角估计及实时生成合成图的方法

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: 17909723

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: 17909723

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 17909723

Country of ref document: EP

Kind code of ref document: A1