WO2017101150A1 - 结构光三维扫描***的标定方法及装置 - Google Patents

结构光三维扫描***的标定方法及装置 Download PDF

Info

Publication number
WO2017101150A1
WO2017101150A1 PCT/CN2015/098936 CN2015098936W WO2017101150A1 WO 2017101150 A1 WO2017101150 A1 WO 2017101150A1 CN 2015098936 W CN2015098936 W CN 2015098936W WO 2017101150 A1 WO2017101150 A1 WO 2017101150A1
Authority
WO
WIPO (PCT)
Prior art keywords
calibration
indicator
plane
dimensional
cloud data
Prior art date
Application number
PCT/CN2015/098936
Other languages
English (en)
French (fr)
Inventor
宋展
叶于平
Original Assignee
深圳先进技术研究院
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 深圳先进技术研究院 filed Critical 深圳先进技术研究院
Publication of WO2017101150A1 publication Critical patent/WO2017101150A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis

Definitions

  • the invention relates to the technical field of three-dimensional scanning system calibration, in particular to a calibration method and device for a structured light three-dimensional scanning system.
  • structured light 3D scanning system based on projector and camera architecture is the most extensive non-contact 3D measurement method.
  • a specific coding optical pattern is projected by a projector, and a projected image is acquired by a camera.
  • the image decoding method is used to acquire the projection feature, and then the matching relationship between the projector and the camera is established, and the three-dimensional reconstruction process is realized.
  • the first key issue involved is the calibration problem of stereo vision system, namely how to obtain the internal parameters and external parameters of the projector, camera, and then, by this parameter Establish a triangulation function to calculate the three-dimensional coordinates.
  • the first one is calibration using a three-dimensional calibration object, which has accurate known three-dimensional information, but the disadvantage of this method is that the three-dimensional calibration material needs special fabrication and high precision. Therefore, the application is less; the second method is to use the calibration checkerboard to mark the corners of the checkerboard.
  • the method includes: first, calibrating the camera, and then projecting a number of projectors according to the calibration result of the camera.
  • the basic strategy of the calibration method of structured light 3D scanning system based on checkerboard is to use the accurately printed checkerboard image and paste it on the surface of a standard plane object. First, put the checkerboard at a certain position and shoot the checkerboard by the camera. Image, extracting the corner information by the corner detection algorithm, and then projecting a checkerboard image by the projector, and then detecting the corner information of the projected checkerboard by the corner detection algorithm; then changing the posture of the checkerboard calibration object or The distance, so many times, is obtained to obtain enough calibration image information (usually 15 pairs of images with different poses or positions).
  • the calibration process is first performed on the camera based on the printed checkerboard image, and then Estimating the spatial plane three-dimensional information of the checkerboard, and then calculating the three-dimensional information of the projected checkerboard corner point on the spatial plane, and using the known corner point information of the projected checkerboard plane on the projector plane, the projector can be completed. Calibration, as well as an estimate of the external parameters of the camera and projector.
  • the Chinese patent application No. CN201410164584 discloses a high-precision projector-camera calibration system and calibration method.
  • the main method of the invention is: using a camera calibration method to calibrate the camera to obtain parameters in the camera;
  • the white pattern is superimposed with the pattern of the calibration plate to capture the image of the calibration area;
  • the corner coordinates in the calibration area image are extracted by correcting the image distortion by using the parameters in the camera; and the correspondence between the camera image plane and the calibration plate plane is estimated according to the correspondence relationship of the corner points.
  • the homography matrix sequentially project different specific checkerboard patterns to the calibration plate and respectively superimpose the patterns of the calibration plate to respectively capture the calibration area image; extract the corner coordinate on the calibration plate plane after the image is differentiated and filtered by the calibration area; After taking the average of the corner coordinates, apply the homography matrix to map the corner points to the calibration plate plane; repeat the above steps according to the acquisition of the corner points, and calibrate the projector by the camera calibration method.
  • This calibration method has been widely used and studied in the industry and will not be described here.
  • the main problems in the calibration method of structured light 3D scanning system based on checkerboard include: 1) The checkerboard needs accurate printing and production to ensure accurate size, and the detection error of the checkerboard corner detection process itself will also be The final calibration result has an effect; 2) the plane that pastes the checkerboard requires a higher degree of flatness, because this calibration method is based on the plane hypothesis; 3) the number of corner points contained in the checkerboard is limited, The coverage is difficult to cover the entire projection range and camera shooting range. The image area without corners is difficult to estimate. 4) The existing calibration method is based on the minimum remapping error of the corner point on the plane.
  • the existing calibration methods mainly include: 1) There are many problems in the calibration process that need attention, such as the production of the calibration plate, the position of the calibration process, the number of calibration images, etc. If the experience is insufficient, It is easy to get poor calibration parameters; 2) calibration parameter optimization is based on the detection of the minimum coordinate error of the mapped 2D image from the corner point to the calibration plane, while the actual measurement is based on the three-dimensional scale such as the three-dimensional coordinate distance. Therefore, the calibration optimization error is difficult to reflect the actual calibration parameters. Therefore, the existing calibration method is used to calibrate the structured light three-dimensional scanning system, and the calibration result is not accurate.
  • Embodiments of the present invention provide a calibration method for a structured light three-dimensional scanning system for improving the accuracy of calibration results, and the method includes:
  • the calibration plane is scanned to obtain three-dimensional point cloud data of the calibration plane; the calibration plane is provided with a plurality of uniformly distributed marker points;
  • the first indicator is: calculating according to the three-dimensional point cloud data corresponding to the marked point Obtaining an average of a distance between two points of each of the adjacent points and an actual distance between the two points
  • the second indicator is: calculating according to the three-dimensional point cloud data corresponding to the points An average value of the angle between the angle between the two straight lines formed by connecting each of the marked points and its adjacent two marked points and the actual angle between the two straight lines
  • the third index The difference between the calculated flatness of the calibration plane and the actual flatness of the calibration plane is calculated according to the three-dimensional point cloud data of the calibration plane.
  • the embodiment of the invention further provides a calibration device for a structured light three-dimensional scanning system for improving the accuracy of the calibration result, the device comprising:
  • a three-dimensional point cloud data acquiring module configured to scan the calibration plane according to the initial calibration parameter to obtain three-dimensional point cloud data of the calibration plane; and the calibration plane is provided with a plurality of uniformly distributed marking points;
  • An optimal calibration parameter calculation module is configured to find an optimal calibration parameter of the structured light three-dimensional scanning system according to the relationship between the first indicator, the second indicator, and the third indicator and the calibration parameter;
  • the first indicator is: according to the label The corresponding three-dimensional point cloud data, the calculated average value of the distance between two adjacent marked points and the actual distance difference between the two marked points;
  • the second indicator is: according to the marking The corresponding three-dimensional point cloud data, the calculated angle between the angle between the two points formed by connecting each marked point and its two adjacent points, and the actual angle difference between the two straight lines
  • the third index is: a difference between the calculated flatness of the calibration plane and the actual flatness of the calibration plane according to the three-dimensional point cloud data of the calibration plane.
  • the technical solution provided by the embodiment of the present invention has at least the following beneficial technical effects:
  • the technical solution of the present invention does not need to obtain sufficient calibration image information as in the prior art, and only needs to obtain initial calibration parameters. Based on the initial calibration parameters, according to the first indicator, the second indicator, and the first The relationship between the three indicators and the calibration parameters can be found to find the optimal calibration parameters of the structured light three-dimensional scanning system, so that the technical solution of the invention is not only efficient and simple, but also accurate and reliable.
  • a calibration plane a plurality of uniformly distributed marker points are arranged on the calibration plane, and the calibration plane is subjected to a one-time complete scan using the initial calibration parameters to obtain three-dimensional point cloud data on the calibration plane; Find the 3D point cloud data corresponding to each marker point in the 3D point cloud data; then, according to the first indicator, The relationship between the second indicator and the third indicator and the calibration parameter finds the optimal calibration parameter of the structured light three-dimensional scanning system.
  • the first indicator and the second indicator are obtained according to the three-dimensional point cloud data corresponding to the marker point, and the third indicator is also based on The three-dimensional point cloud data of the calibration plane is calculated, that is, the process of finding the optimal calibration parameters is performed in three-dimensional space, which can improve the accuracy of the calibration result; in addition, the most accurate three-dimensional scanning system for structural light is sought.
  • the optimal calibration parameters take into account the following three indicators: the first indicator: the average of the difference between the distance between two points and the actual distance between the two points; the second indicator: each point The angle between the angle between the two straight lines formed by the two adjacent points connected to it and the actual angle difference between the two straight lines; and the third index: the flatness of the calibration plane The difference in the actual flatness of the calibration plane ensures that the optimal system calibration parameters are obtained.
  • the technical solution provided by the embodiment of the present invention improves the accuracy of the calibration result, thereby improving the three-dimensional measurement accuracy and reliability of the structured light three-dimensional scanning system.
  • FIG. 1 is a schematic flow chart of a calibration method of a structured light three-dimensional scanning system according to an embodiment of the present invention
  • Figure 2 is a schematic illustration of a calibration plane used in an embodiment of the present invention.
  • FIG. 3 is a schematic structural view of a calibration device for a structured light three-dimensional scanning system according to an embodiment of the present invention.
  • the present invention proposes a calibration method for a structured light three-dimensional scanning system, and the optimization target used is a standard plane (such as a glass whiteboard), and a plurality of marking points are marked in advance on the plane, and the marking is performed. The size between the points is accurately printed.
  • the three-dimensional point cloud data is obtained through the initial calibration parameters, and the relationship between the first index, the second index and the third index and the calibration parameters is obtained.
  • the optimal calibration parameters of the structured light 3D scanning system are found.
  • the whole process is simple and easy.
  • the obtained optimal calibration parameters can ensure the improvement of the measurement accuracy of the whole scanning system.
  • the result is stable and reliable, and the accuracy of the initial calibration parameters.
  • the requirements are not high and can be widely applied to the existing structural light system calibration process to improve the accuracy and stability of the calibration results. The details will be described below.
  • FIG. 1 is a schematic flow chart of a calibration method of a structured light three-dimensional scanning system according to an embodiment of the present invention; as shown in FIG. 1, the method includes the following steps:
  • Step 101 Scan the calibration plane according to the initial calibration parameter to obtain three-dimensional point cloud data of the calibration plane; and set a plurality of uniformly distributed marker points on the calibration plane;
  • Step 102 Find an optimal calibration parameter of the structured light three-dimensional scanning system according to the relationship between the first indicator, the second indicator, and the third indicator and the calibration parameter;
  • the first indicator is: a three-dimensional point cloud corresponding to the marked point Data, the calculated average value of the distance between two adjacent points of each adjacent point and the actual distance between the two points;
  • the second indicator is: a three-dimensional point cloud corresponding to the point Data, an average value of the angle between the angle between the calculated two points and the two lines formed by the adjacent two points, and the actual angle difference between the two lines;
  • the third indicator is: a difference between the calculated flatness of the calibration plane and the actual flatness of the calibration plane according to the three-dimensional point cloud data of the calibration plane.
  • an initial calibration parameter of the structured light three-dimensional scanning system is obtained, which can be implemented by the following steps:
  • M image coordinates, M-three-dimensional coordinates, (M is relative to the camera, Mp relative to the projector);
  • Mc (p) [X c(p) , Y c(p) , Z c(p) ] T to represent the three-dimensional coordinates of a point on the object, where subscript c(p), c represents Relative to the camera (camera) coordinate system, p represents the relative to the projector coordinate system.
  • K is the internal parameter matrix of the camera/projector, ie:
  • f c(p) x, f c(p) y which represents the pixel scale factor in the xy direction
  • r represents the angular distortion factor of the pixel plane XY axis
  • c c(p) x, c c(p) y which represents the center point of the image coordinate.
  • the above formula describes the model and parameters of the camera-projector in the basic structured light system.
  • the position of the three-dimensional space point can be obtained by using the trigonometric principle for the corresponding point to be found:
  • equation (5) we can calculate its X, Y coordinates.
  • a complete scan of the calibration plane is completed.
  • the three-dimensional point cloud data of the calibration plane can be acquired; the image coordinates of all the markers in FIG. 2 are detected from the scanned image, and found in the three-dimensional scan data.
  • the initial calibration parameter is taken as the initial value of the global optimization function, and the upper and lower limit range thresholds of the relevant calibration parameters are given, and the global optimization function is executed until the error index is minimized by one or any combination of all the error indicators, so that the optimal calibration parameter can be obtained.
  • the optimal calibration parameter of the structured light three-dimensional scanning system is found, and the global optimization function may be established by using the above-mentioned method.
  • the method is obtained, for example:
  • the technical solution provided by the embodiment of the present invention does not need to obtain sufficient calibration image information as in the prior art, and only needs to obtain initial calibration parameters, and based on the initial calibration parameters, according to the established global optimization function. Further calibration parameter optimization, find one of the three indicators or the combination of the minimum calibration parameters as the optimal calibration parameters, so that the technical solution of the present invention is not only efficient and simple, but also accurate and reliable.
  • the foregoing “using the initial calibration parameter as an initial value of the global optimization function, and iteratively finding an optimal calibration parameter” may include:
  • the advantages of the upper and lower limits of the set optimization calibration parameter range are: it can effectively avoid the optimization to enter the local minimum and improve the operation efficiency.
  • the calibration plane may be constructed in various manners.
  • the calibration plane may include: a planar object and a surface object that is adhered to the planar object and satisfies the scanning range of the structured light three-dimensional scanning system.
  • White paper the white paper is provided with a plurality of evenly distributed marking points.
  • the manufacturing process may be: designing a white grid point image according to the scanning range of the structured light system to be calibrated, and printing and pasting on the plane object, the main purpose of which is to mark the white surface.
  • the calibration plane can also be a specially made glass whiteboard provided with a plurality of evenly distributed marking points.
  • the shape of the marking point may be a circle.
  • the advantage that the shape of the marker point can be a circle is: when calculating the above three indicators, taking the distance between the two marker points as an example, the distance between the center points of the two marker points needs to be calculated. In the same way, when calculating the other two indicators, it is also calculated based on the center point.
  • the center point of the circle is the center of the circle, and the inside of the three-dimensional space is the center of the circle, which is convenient for finding the center point.
  • other shapes of marking points can also be selected, as long as it is convenient to accurately calculate three indicators.
  • the calibration parameters mentioned in the embodiments of the present invention may include internal parameters of the camera and the camera and all external parameters in the structured light three-dimensional scanning system.
  • the global optimization function in the embodiment of the present invention includes all the internal parameters of the camera and the projector, and 12 external parameters (9 rotation matrices, 3 translation vectors), and the optimization objective function minimization criterion includes three indicators. The following three indicators are described in detail:
  • the specific calculation process of flatness can be: due to the huge number of reconstructed point clouds, up to several million, in order to improve the running speed of the optimization function, the specific implementation can be calculated according to the three-dimensional point cloud data uniformly sampled from the calibration plane.
  • the difference between the obtained flatness of the calibration plane and the actual flatness of the calibration plane may be: random sampling or uniform sampling of the point cloud to obtain sparse point cloud data, and least squares the point cloud data. Plane fitting, obtaining the flatness error E1 of the calibration plane;
  • the specific calculation process can be: since the size of the marked points on the calibration plane is precisely known (the dimensional data designed when the calibration plane is made), we calculate the distance error between all the marked points, take the average, and calculate The error between the distance value and the true distance value E2;
  • the difference between the cosine value of the angle (calculated according to the three-dimensional point cloud data) and the standard angle (the actual angle of the angle, such as 90 degrees) can be used as the angle error value:
  • E3
  • satisfy: E d ⁇ E p ⁇ E a ;
  • E d is the first index (size error): the distance between two marked points per phase and the two marked points The average value of the actual distance difference between the two;
  • E a is the second index (angle error): according to the three-dimensional point cloud data corresponding to the marked point, each calculated point is connected with two adjacent points of the mark The average angle difference between the angle between the two straight lines and the actual angle between the two straight lines;
  • E p is the third index (flatness error): calculated according to the 3D point cloud data of the calibration plane The difference between the obtained flatness of the calibration plane and the actual flatness of the calibration plane.
  • the global optimization function (8) is a typical nonlinear, multi-objective optimization problem, which can be performed by using existing optimization tools (for example, MATLAB tools). Optimize and finally obtain the optimal calibration parameters that meet the three optimization criteria (three indicators).
  • three indicators we comprehensively consider three indicators, specific implementation, or select individual or arbitrary indicators for optimization according to actual needs.
  • the technical solution provided by the implementation of the present invention does not need to obtain sufficient calibration image information as in the prior art, and only needs to obtain initial calibration parameters, and based on the initial calibration parameters, according to the established global optimization.
  • the function performs further calibration parameter optimization, and finds the calibration parameter corresponding to one or more of the three indicators as the optimal calibration parameter, so that the technical solution of the invention is not only efficient and simple, but also accurate and reliable;
  • the technical solution provided by the implementation of the invention provides a calibration plane by setting a plurality of uniformly distributed marking points on the calibration plane, and performing a one-time complete scanning on the calibration plane by using initial calibration parameters to obtain all points on the calibration plane.
  • 3D point cloud data then, the 3D point cloud data corresponding to each marker point is found in the 3D point cloud data; then, a global optimization function is established, the initial calibration parameter is the initial value of the global optimization function, and the iterative calculation finds three indicators.
  • One or more minimum time corresponding calibration parameters are used as the optimal calibration parameters.
  • the calculation of the three indicators is based on the three-dimensional point cloud data of all points on the calibration plane and the three-dimensional point cloud data corresponding to each marked point, that is, the process of optimizing the calibration parameters is performed in three-dimensional space, so that Improve the accuracy of the calibration results;
  • the objective function of the global optimization function includes one or any combination of the following three key geometric attribute indicators: the flatness of the calibration plane, the distance between each two adjacent points and the location The difference between the actual distance between the two marked points, the angle between the two straight lines formed by the connection of each marked point and its adjacent two marked points, and the actual angle between the two straight lines Differences, this ensures that the optimal system calibration parameters are obtained.
  • the technical solution provided by the embodiment of the present invention improves the accuracy of the calibration result, thereby improving the three-dimensional measurement accuracy and reliability of the structured light three-dimensional scanning system.
  • an embodiment of the present invention further provides a calibration device for a structured light three-dimensional scanning system, such as the following embodiment. Since the principle of solving the problem of the calibration device of the structured light three-dimensional scanning system is similar to the calibration method of the structured light three-dimensional scanning system, the implementation of the calibration device of the structured light three-dimensional scanning system can be referred to the implementation of the calibration method of the structured light three-dimensional scanning system, and repeated I won't go into details here.
  • the terms "unit”, “device” or “module” may implement a combination of software and/or hardware of a predetermined function.
  • the apparatus described in the following embodiments may be implemented in software, hardware, or a combination of software and hardware, and is also possible and conceivable.
  • FIG. 3 is a schematic structural diagram of a calibration apparatus for a structured light three-dimensional scanning system according to an embodiment of the present invention. As shown in FIG. 3, the apparatus includes:
  • the three-dimensional point cloud data acquiring module 10 is configured to scan the calibration plane according to the initial calibration parameter to obtain three-dimensional point cloud data of the calibration plane; and the calibration plane is provided with a plurality of uniformly distributed marker points;
  • the optimal calibration parameter calculation module 20 is configured to find an optimal calibration parameter of the structured light three-dimensional scanning system according to the relationship between the first indicator, the second indicator, and the third indicator and the calibration parameter;
  • the first indicator is: according to the The three-dimensional point cloud data corresponding to the marked point, the calculated average value of the distance between two adjacent marked points and the actual distance between the two marked points;
  • the second indicator is: according to the The three-dimensional point cloud data corresponding to the marked point, the calculated angle between the angle between each of the marked points and the two straight lines formed by the adjacent two marked points and the actual angle between the two straight lines
  • the average value of the difference is: the difference between the calculated flatness of the calibration plane and the actual flatness of the calibration plane according to the three-dimensional point cloud data of the calibration plane.
  • the optimal calibration parameter calculation module is specifically configured to:
  • a global parameter optimization algorithm is proposed. Based on the given calibration plane and marker parameters, a global optimization function including three objective functions of flatness, size and angle is established. Optimization of all parameters.
  • the optimization method of the present invention is performed in the measured three-dimensional space, and the optimization parameters include all system parameters to ensure that the optimal system calibration parameters can be obtained. Minimize system measurement errors.
  • modules, devices or steps of the embodiments of the invention described above may be implemented by a general-purpose computing device, which may be centralized on a single computing device or distributed across multiple computing devices. Alternatively, they may be implemented by program code executable by the computing device such that they may be stored in the storage device by the computing device and, in some cases, may be different.
  • the steps shown or described herein are performed sequentially, or they are separately fabricated into individual integrated circuit modules, or a plurality of modules or steps thereof are fabricated as a single integrated circuit module.
  • embodiments of the invention are not limited to any specific combination of hardware and software.

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

一种结构光三维扫描***的标定方法及装置,该方法包括:根据初始标定参数,对标定平面扫描,得到标定平面的三维点云数据;标定平面上设置有多个均匀分布的标记点(101);根据第一指标、第二指标和第三指标与标定参数的关系,找到最优标定参数(102);第一指标为:根据标记点对应的三维点云数据,计算得到的每相临两个标记点之间的距离与实际距离差异的平均值;第二指标为:根据标记点对应的三维点云数据,计算得到的每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度与实际夹角角度差异的平均值;第三指标为:根据标定平面的三维点云数据,计算得到的标定平面的平面度与实际平面度的差异。

Description

结构光三维扫描***的标定方法及装置
本申请要求2015年12月14日递交的申请号为201510925237.5、发明名称为“结构光三维扫描***的标定方法及装置”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本发明涉及三维扫描***标定技术领域,特别涉及一种结构光三维扫描***的标定方法及装置。
背景技术
目前,基于投影仪和相机架构的结构光三维扫描***是最为广泛的非接触式三维测量手段,该技术基于立体视觉中的三角测量原理,通过投影仪投射特定编码光学图案,由相机获取投射图像,通过图像解码方法获取投射特征,进而建立起投影仪与相机的匹配关系,实现三维重建过程。
基于投影仪和相机架构的结构光三维扫描***进行三维重建,涉及到的首个关键问题就是立体视觉***的标定问题,即如何获取投影仪、相机的内部参数及外部参数,进而,由该参数建立三角测量函数,实现三维坐标的计算。目前常用的标定策略大概有两种:第一种是利用三维标定物进行标定,该标定物具有精确的已知三维信息,但该方法的缺点是三维标定物需要特殊制作,且精度要求很高,因此应用较少;第二种方法是用标定棋盘格,以棋盘格的角点为标定特征,该方法包括:首先,对相机进行标定,然后,根据相机标定结果,由投影仪投射出若干特征点到棋盘格平面上,计算出投射的特征点的三维信息,再对投影仪进行标定,因为这种方法比较易于实现,通过打印的棋盘格贴在平面物体上即可操作,因而为业内广泛使用。
基于棋盘格的结构光三维扫描***标定方法的基本策略是:使用精确打印的棋盘格图像,将其粘贴在一个标准平面物体表面,首先,将棋盘格放在某个位置,由相机拍摄棋盘格图像,通过角点检测算法提取出角点信息,进而由投影仪投射出一个棋盘格图像,再通过角点检测算法检测出投射的棋盘格的角点信息;然后改变棋盘格标定物的姿态或距离,如此重复多次,获取足够多的标定图像信息(常用需15个不同姿态或位置的图像对)。在标定过程中,首先基于打印棋盘格图像对相机完成标定过程,之后就可以 估算出棋盘格的空间平面三维信息,进而通过对投射棋盘格角点在空间平面上的三维信息进行计算,利用投射棋盘格在投影仪平面上的已知角点信息,即可完成对投影仪的标定,以及相机与投影仪的外部参数的估计。
中国专利申请号为CN201410164584的申请即公开了一种高精度的投影仪-摄像机标定***及标定方法,该发明主要的方法是:用摄像机标定方法标定摄像机,得到摄像机内参数;向标定板投影纯白图案并与标定板的图案叠加,捕获标定区域图像;利用摄像机内参数对图像畸变校正后提取标定区域图像中的角点坐标;根据角点的对应关系估计摄像机像平面和标定板平面之间的单应性矩阵;依序向标定板投影不同特定棋盘图案并分别与标定板的图案叠加,分别捕获标定区域图像;对标定区域图像差分、滤波处理后提取标定板平面上的角点坐标;取角点坐标平均值后应用单应性矩阵映射角点至标定板平面;按角点的获取情况重复上述步骤,利用摄像机标定方法标定投影仪。这种标定方法已为业内广泛使用和研究,在此不再赘述。
然而,基于棋盘格的结构光三维扫描***标定方法存在的主要问题包括:1)棋盘格需要精确的打印和制作,保证精确的尺寸,且棋盘格角点检测过程本身存在的检测误差也会对最终的标定结果产生影响;2)粘贴棋盘格的平面需要较高的平面度,因为这种标定方法是建立在平面假设基础之上的;3)棋盘格所包含的角点数量是有限的,其覆盖范围很难囊括整个投影范围和相机拍摄范围,没有角点存在的图像区域,其畸变参数等是难以估计的;4)现有标定方法是以角点在平面上的重映射误差最小为最终优化准则的,在实际测量中,这种误差准则与三角测量的结果存在不一致性,即使重投影误差很小也难以保证最终测量结果达到最优误差;5)标定过程和操作过程存在的人为操作误差导致结果的不确定性增加。
综上所述,现有标定方法存在的问题主要包括:1)标定过程有诸多些问题需要注意,如标定板的制作,标定过程摆放的位置,标定图像的数量等等,如果经验不足,很容易得到较差的标定参数;2)标定参数优化是基于检测到角点到标定平面的映射2D图像坐标误差最小为准则的,而实际测量中是以三维坐标距离等三维尺度为标准的,因此标定优化误差很难反映出实际的标定参数优劣。因此,利用现有标定方法对结构光三维扫描***标定,标定结果不准确。
发明内容
本发明实施例提供了一种结构光三维扫描***的标定方法,用以提高标定结果的准确性,该方法包括:
根据初始标定参数,对标定平面扫描,得到标定平面的三维点云数据;所述标定平面上设置有多个均匀分布的标记点;
根据第一指标、第二指标和第三指标与标定参数的关系,找到结构光三维扫描***的最优标定参数;所述第一指标为:根据所述标记点对应的三维点云数据,计算得到的每相临两个标记点之间的距离与所述两个标记点之间的实际距离差异的平均值;所述第二指标为:根据所述标记点对应的三维点云数据,计算得到的每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度与所述两条直线之间的实际夹角角度差异的平均值;所述第三指标为:根据标定平面的三维点云数据,计算得到的标定平面的平面度与所述标定平面的实际平面度的差异。
本发明实施例还提供了一种结构光三维扫描***的标定装置,用以提高标定结果的准确性,该装置包括:
三维点云数据获取模块,用于根据初始标定参数,对标定平面扫描,得到标定平面的三维点云数据;所述标定平面上设置有多个均匀分布的标记点;
最优标定参数计算模块,用于根据第一指标、第二指标和第三指标与标定参数的关系,找到结构光三维扫描***的最优标定参数;所述第一指标为:根据所述标记点对应的三维点云数据,计算得到的每相临两个标记点之间的距离与所述两个标记点之间的实际距离差异的平均值;所述第二指标为:根据所述标记点对应的三维点云数据,计算得到的每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度与所述两条直线之间的实际夹角角度差异的平均值;所述第三指标为:根据标定平面的三维点云数据,计算得到的标定平面的平面度与所述标定平面的实际平面度的差异。
与现有技术相比较,本发明实施例提供的技术方案至少具有以下有益技术效果:
首先,本发明技术方案,不需要像现有技术中获取足够多的标定图像信息,仅需获得初始的标定参数,在该初始的标定参数的基础上,根据第一指标、第二指标和第三指标与标定参数的关系,找到结构光三维扫描***的最优标定参数即可,这样本发明技术方案不仅实施高效简单,而且标定结果准确可靠;
其次,通过建立一个标定平面,该标定平面上设置有多个均匀分布的标记点,采用初始标定参数,对该标定平面进行一次性完整扫描,得到标定平面上的三维点云数据;然后,在三维点云数据中找到每个标记点对应的三维点云数据;接着,根据第一指标、 第二指标和第三指标与标定参数的关系,找到结构光三维扫描***的最优标定参数,第一指标和第二指标是根据标记点对应的三维点云数据得到的,第三指标也是根据标定平面的三维点云数据进行计算得到的,也就是说,寻找最优标定参数的过程是在三维空间中进行的,这样可以提高标定结果的准确性;另外,寻找结构光三维扫描***的最优标定参数考虑了如下三个指标:第一指标:每相临两个标记点之间的距离与所述两个标记点之间的实际距离差异的平均值;第二指标:每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度与所述两条直线之间的实际夹角角度差异的平均值;以及第三指标:标定平面的平面度与所述标定平面的实际平面度的差异,这样保证了获取最优的***标定参数。
通过上述可知,本发明实施例提供的技术方案,提高了标定结果的准确性,进而提高了结构光三维扫描***的三维测量精度和可靠性。
附图说明
为了更清楚地说明本发明实施例中的技术方案,下面将对实施例描述中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本发明的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。在附图中:
图1是本发明实施例中结构光三维扫描***的标定方法的流程示意图;
图2是本发明实施例中使用的标定平面的示意图;
图3是本发明实施例中结构光三维扫描***的标定装置的结构示意图。
具体实施方式
为使本发明实施例的目的、技术方案和优点更加清楚明白,下面结合附图对本发明实施例做进一步详细说明。在此,本发明的示意性实施例及其说明用于解释本发明,但并不作为对本发明的限定。
在实际工作中,发明人发现现有结构光三维扫描***的标定方法存在的问题主要有两点:1)标定过程有诸多些问题需要注意,如标定板的制作,标定过程摆放的位置,标定图像的数量等等,如果经验不足,很容易得到较差的标定参数;2)现有方法的标定参数优化是基于检测到角点到标定平面的映射2D图像坐标误差最小为准则的,而实际测 量中是以三维坐标距离等三维尺度为标准的,因此标定优化误差很难反映出实际的标定参数优劣。
由于发明人发现了上述技术问题,本发明提出了一种结构光三维扫描***的标定方法,所使用的优化目标为一标准平面(如玻璃白板),平面上预先标记出若干个标记点,标记点之间的尺寸为精确打印所得,通过对该标定板的一次性3D扫描,通过初始标定参数,获取其三维点云数据,通过第一指标、第二指标和第三指标与标定参数的关系,找到结构光三维扫描***的最优标定参数,整个过程简单易行,得到的最优标定参数能保证整个扫描***在测量精度上有较大的提升,结果稳定可靠,对初始标定参数的精度要求不高,可以广泛应用于现有的结构光***标定过程,用于提高标定结果的准确性和稳定性。下面进行详细说明。
图1是本发明实施例中结构光三维扫描***的标定方法的流程示意图;如图1所示,该方法包括如下步骤:
步骤101:根据初始标定参数,对标定平面扫描,得到标定平面的三维点云数据;所述标定平面上设置有多个均匀分布的标记点;
步骤102:根据第一指标、第二指标和第三指标与标定参数的关系,找到结构光三维扫描***的最优标定参数;所述第一指标为:根据所述标记点对应的三维点云数据,计算得到的每相临两个标记点之间的距离与所述两个标记点之间的实际距离差异的平均值;所述第二指标为:根据所述标记点对应的三维点云数据,计算得到的每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度与所述两条直线之间的实际夹角角度差异的平均值;所述第三指标为:根据标定平面的三维点云数据,计算得到的标定平面的平面度与所述标定平面的实际平面度的差异。
具体实施时,在上述步骤101中,首先要获取结构光三维扫描***的初始标定参数,其可以通过如下步骤实现:
参数定义如下:
c-相机,p-投影仪;
m—图像坐标,M-三维坐标,(相对于相机为Mc,相对于投影仪为Mp);
k-径向畸变,R,T—相机投影仪之间旋转,平移参数。
用Mc(p)=[Xc(p),Yc(p),Zc(p)]T表示物体上某个点的三维空间坐标,其中下标c(p),c代表的是相对于摄像机(相机)坐标系,p代表的是相对于投影仪坐标系。
那么,我们根据投射投影模型,可以将其图像坐标表示为
Figure PCTCN2015098936-appb-000001
Figure PCTCN2015098936-appb-000002
在考虑了径向畸变和切向误差之后有,对
Figure PCTCN2015098936-appb-000003
图像坐标进行反畸变处理,
Figure PCTCN2015098936-appb-000004
其中,
Figure PCTCN2015098936-appb-000005
而式子(2)中,
Figure PCTCN2015098936-appb-000006
代表的是切向畸变误差。
Figure PCTCN2015098936-appb-000007
那么,最后可以得到:
Figure PCTCN2015098936-appb-000008
其中,K是相机/投影仪的内部参数矩阵,即:
fc(p)x,fc(p)y,表示x-y方向像素尺度因子,r表示像素平面X-Y轴夹角畸变因子,cc(p)x,cc(p)y,表示图像中心点坐标。
Figure PCTCN2015098936-appb-000009
上文中说过我们还标定摄像机和投影仪的相对位置,那么有:
Figure PCTCN2015098936-appb-000010
上述公式描述了基本的结构光***中,相机-投影仪的模型和参数,在本发明中,我们首先利用传统的摄像机标定方法进行相机标定,之后通过利用摄像机去捕捉投影仪投射棋盘图案到平面上的图案。建立起投影仪坐标和摄像机坐标系之间的一个单应矩阵,将投影仪建立与相机一样的模型,我们可以得到相机-投影仪的内部、外部初始标定参数。
在得到结构光***的标定结果之后,对需找到的对应点利用三角原理可以得到其三维空间点的位置是:
Figure PCTCN2015098936-appb-000011
再利用式子(5)我们可以计算出其X,Y坐标。
基于上述方法,我们可以得到相机-投影仪的内部及外部参数,及实现每个图像点的三维坐标计算,在此基础上,我们完成一次对标定平面(图2)的扫描过程,得到高密度的点云数据和图2中标记点的中心点(本发明提到的计算三个指标时,都是根据该标记点的中心点的坐标来计算的)图像坐标及三维坐标,在此基础上,我们可以寻找最优标定参数,例如:标定参数的优化,即在获取初始标定参数之后,就可以寻找最优标定参数,例如:进行标定参数的优化,其大致过程可以如下:
采用初始标定参数,完成对标定平面的一次完整扫描,此时,可以获取该标定平面的三维点云数据;从扫描图像中检测出图2中所有标记点的图像坐标,在三维扫描数据中找到标记点的图像坐标对应的三维坐标;建立全局优化函数,该函数包括相机和投影机的内部参数以及全部外部参数,其优化指标包括三个:平面度:基于所有重建点的平面拟合误差,即根据标定平面上三维点云数据计算出的平面度与所述标定平面的实际平面度的差异;尺寸精度:检测出的标记点的尺寸误差,即每两个标记点之间的距离与实际距离的差异;角度误差:检测出的标记点的夹角误差,即每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度与实际夹角角度的差异。最后,以初始标定参数为全局优化函数初始值,给定相关标定参数的上下限范围阈值,执行全局优化函数,直至所有误差指标其中之一或任意组合最小时最小,即可获得最优标定参数。
下面详细介绍上述优化标定参数的步骤:
具体实施时,本发明实施例中,根据第一指标、第二指标和第三指标与标定参数的关系,找到结构光三维扫描***的最优标定参数,可以通过上述提到的建立全局优化函数的方法得到,例如:
建立所述第一指标、第二指标和第三指标与标定参数的全局优化函数;
以所述初始标定参数为所述全局优化函数的初始值,迭代找到最优标定参数,对于每个迭代周期均执行以下操作:
根据所述标记点对应的三维点云数据,计算所述第一指标和第二指标;根据标定平面的三维点云数据,计算所述第三指标;直到找到第一指标、第二指标和第三指标中一个或多个最小时对应的标定参数,作为结构光三维扫描***的最优标定参数。
本发明实施例提供的技术方案,不需要像现有技术中获取足够多的标定图像信息,仅需获得初始的标定参数,在该初始的标定参数的基础上,根据建立的全局优化函数,进行进一步的标定参数优化,找到三个指标的其中之一或任意组合最小时对应的标定参数,作为最优标定参数,这样本发明技术方案不仅实施高效简单,而且标定结果准确可靠。
当然,除了通过建立第一指标、第二指标和第三指标与标定参数的全局优化函数,来迭代找到最优标定参数的方式之外,还可以有多种方式,例如,可以通过建立一个第一指标、第二指标和第三指标与标定参数的表格等等方式,来找到最优标定参数。
具体实施时,上述“以所述初始标定参数为所述全局优化函数的初始值,迭代找到最优标定参数”可以包括:
以所述初始标定参数为全局优化函数的初始值,设定标定参数的范围阈值,迭代找到最优标定参数。
在进行优化的过程中,在初始标定参数基础上,设定的优化标定参数范围的上下限的优点是:可以有效避免优化进入局部最小,同时提高运行效率。
具体实施时,在上述步骤101中,标定平面的构建可以有多种方式,例如图2所示,标定平面可以包括:平面物体以及粘贴在平面物体上的、满足结构光三维扫描***扫描范围的白纸;所述白纸上设置有多个均匀分布的标记点。
具体实施时,制作过程可以为:根据所需标定的结构光***的扫描范围设计一张白色的网格点图像,将其打印后粘贴与平面物体上,其主要目的是在白色表面上标记出若干精确尺寸参考点。当然,标定平面也可以是特意制作的、设置有多个均匀分布的标记点的玻璃白板。
具体实施时,所述标记点的形状可以为圆形。标记点的形状可以为圆形的优点是:在计算上述提到的三个指标时,以计算两个标记点之间的距离为例,需要计算两个标记点的中心点之间的距离,同理,在计算其它两个指标时,也是以中心点为标准进行计算的,圆形的中心点就是圆心,三维空间里面是球心,方便寻找中心点。实际实施中,也可选择其它形状的标记点,只要方便精确计算三个指标即可。
本发明实施例提到的标定参数可以包括:结构光三维扫描***中相机和摄影机的内部参数和全部外部参数。本发明实施例中的全局优化函数包含了所有的相机、投影仪的内部参数,以及12个外部参数(旋转矩阵9个,平移向量3个),其优化目标函数最小化准则包含三个指标,下面对这三个指标进行详细介绍:
a)平面度误差指标E1(即为:根据标定平面上所有点的三维点云数据,计算出的标定平面的平面度与所述标定平面的实际平面度的差异):
平面度具体的计算过程可以为:由于重建的点云数量巨大,可达几百万,为了提高优化函数的运行速度,具体实施时,可以根据从标定平面上均匀采样的三维点云数据,计算得到的标定平面的平面度与所述标定平面的实际平面度的差异,即可以为:对点云采用随机采样或者均匀采样,得到较为稀疏的点云数据,对这些点云数据进行最小二乘平面拟合,得到标定平面的平面度误差E1;
b)尺寸误差指标E2(即为:根据每个标记点对应的三维点云数据,计算出的每相临两个标记点之间的距离与所述两个标记点之间的实际距离差异的平均值):
具体的计算过程可以为:由于标定平面上的标记点的尺寸为精确已知(在制作标定平面时设计的尺寸数据),因此,我们计算所有标记点之间的距离误差,取平均值,计算的距离值与真实距离值的误差E2;
c)角度误差E3(即为:每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度与所述两条直线之间的实际夹角角度差异的平均值):
具体实施时,可以采用角度(根据三维点云数据,计算得到的)的余弦值与标准的角度(实际的夹角角度,如90度)进行比较后的差异,作为角度误差值:
E3=∑|cosθcal-cosθreal|,其中,θcal为根据三维点云数据,计算得到的每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度,θreal为所述两条直线之间的实际夹角角度(即,标定平面精确设计打印时的标准尺寸)。
基于上述三个误差标准,我们就可以构造出全局优化函数的目标函数如下:
Fobj=∑(Ed+α·Ep+β·Ea);  (7)
其中,α,β满足:Ed≈α·Ep≈β·Ea;Ed为第一指标(尺寸误差):每相临两个标记点之间的距离与所述两个标记点之间的实际距离差异的平均值;Ea为第二指标(角度误差):根据所述标记点对应的三维点云数据,计算得到的每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度与所述两条直线之间的实际夹角角度差异的平均值;Ep为第三指标(平面度误差):根据标定平面的三维点云数据,计算得到的标定平面的平面度与所述标定平面的实际平面度的差异。α和β是经验参数,用来将三个误差调整到一个数量级,比如尺寸误差是0.1mm左右的话,平面度误差常在1mm左右,角度误差在0.1度左右的话,则可设置为α=0.1和β=1。
最后,我们可以建立全局优化函数如下:
minFobj(parasSLS)
s.t.  (8)
Figure PCTCN2015098936-appb-000012
其中,
Figure PCTCN2015098936-appb-000013
是在初始标定参数基础上,设定的标定参数的上限,
Figure PCTCN2015098936-appb-000014
是在初始标定参数基础上,设定的标定参数的下限,这样可以有效避免优化进入局部最小,同时提高运行效率。
具体实施时,在对上述全局优化函数(8)求解时:全局优化函数(8)属于典型的非线性、多目标优化问题,我们可以借助现有的优化工具(例如:MATLAB工具)对其进行优化,并最终获取能够满足三个优化准则(三个指标)对应的最优标定参数。在所建立的全局优化函数中,我们综合考虑了三个指标,具体实施,亦可根据实际需要,选择单个或任意个指标进行优化处理。
通过上述可知,本发明实施提供的技术方案,不需要像现有技术中获取足够多的标定图像信息,仅需获得初始的标定参数,在该初始的标定参数的基础上,根据建立的全局优化函数,进行进一步的标定参数优化,找到三个指标中一个或多个最小时对应的标定参数,作为最优标定参数,这样本发明技术方案不仅实施高效简单,而且标定结果准确可靠;
本发明实施提供的技术方案,通过建立一个标定平面,该标定平面上设置有多个均匀分布的标记点,采用初始标定参数,对该标定平面进行一次性完整扫描,得到标定平面上所有点的三维点云数据;然后,在三维点云数据中找到每个标记点对应的三维点云数据;接着,建立全局优化函数,以初始标定参数为全局优化函数的初始值,迭代计算找到三个指标的一个或多个最小时对应的标定参数,作为最优标定参数。计算三个指标是根据标定平面上所有点的三维点云数据和每个标记点对应的三维点云数据进行计算的,也就是说,优化标定参数的过程是在三维空间中进行的,这样可以提高标定结果的准确性;另外,全局优化函数的目标函数包括以下三个关键几何属性指标的其中之一或任意组合:标定平面的平面度,每相临两个标记点之间的距离与所述两个标记点之间的实际距离差异、每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度与所述两条直线之间的实际夹角角度差异,这样保证了获取最优的***标定参数。
通过上述可知,本发明实施例提供的技术方案,提高了标定结果的准确性,进而提高了结构光三维扫描***的三维测量精度和可靠性。
基于同一发明构思,本发明实施例还提供了一种结构光三维扫描***的标定装置,如下面的实施例。由于结构光三维扫描***的标定装置解决问题的原理与结构光三维扫描***的标定方法相似,因此结构光三维扫描***的标定装置的实施可以参见结构光三维扫描***的标定方法的实施,重复之处不再赘述。以下所使用的,术语“单元”、“装置”或者“模块”可以实现预定功能的软件和/或硬件的组合。以下实施例所描述的装置可以以软件、硬件,或者软件和硬件的组合的实现,也是可能并被构想的。
图3是本发明实施例中结构光三维扫描***的标定装置的结构示意图,如图3所示,该装置包括:
三维点云数据获取模块10,用于根据初始标定参数,对标定平面扫描,得到标定平面的三维点云数据;所述标定平面上设置有多个均匀分布的标记点;
最优标定参数计算模块20,用于根据第一指标、第二指标和第三指标与标定参数的关系,找到结构光三维扫描***的最优标定参数;所述第一指标为:根据所述标记点对应的三维点云数据,计算得到的每相临两个标记点之间的距离与所述两个标记点之间的实际距离差异的平均值;所述第二指标为:根据所述标记点对应的三维点云数据,计算得到的每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度与所述两条直线之间的实际夹角角度差异的平均值;所述第三指标为:根据标定平面的三维点云数据,计算得到的标定平面的平面度与所述标定平面的实际平面度的差异。
在一个实施例中,所述最优标定参数计算模块具体用于:
建立所述第一指标、第二指标和第三指标与标定参数的全局优化函数;
以所述初始标定参数为所述全局优化函数的初始值,迭代找到最优标定参数,对于每个迭代周期均执行以下操作:
根据所述标记点对应的三维点云数据,计算所述第一指标和第二指标;根据标定平面的三维点云数据,计算所述第三指标;直到找到第一指标、第二指标和第三指标中一个或多个最小时对应的标定参数,作为结构光三维扫描***的最优标定参数。
经过发明人大量的实验认证,本发明实施例效果很理想,和设计的预期一致。本发明实施例提供的技术方案的有益技术效果为:
1)使用了一个简单的带有若干尺寸标记点的标定平面作为结构光三维扫描***标定参数优化的实验目标,实施简单,结果可靠。
2)提出了一种全局参数优化算法,基于给定的标定平面和标记点参数,建立了包含平面度、尺寸、角度三个目标函数的全局优化函数,通过一次平面扫描数据,即可完成对全部参数的优化过程。
3)与现有的标定参数优化方面不同的是,本发明的优化方法是在实测的三维空间中进行的,而且其优化参数包含了所有的***参数,确保能够获取最优的***标定参数,将***测量误差降到最低。
显然,本领域的技术人员应该明白,上述的本发明实施例的各模块、装置或各步骤可以用通用的计算装置来实现,它们可以集中在单个的计算装置上,或者分布在多个计算装置所组成的网络上,可选地,它们可以用计算装置可执行的程序代码来实现,从而,可以将它们存储在存储装置中由计算装置来执行,并且在某些情况下,可以以不同于此处的顺序执行所示出或描述的步骤,或者将它们分别制作成各个集成电路模块,或者将它们中的多个模块或步骤制作成单个集成电路模块来实现。这样,本发明实施例不限制于任何特定的硬件和软件结合。
以上仅为本发明的优选实施例而已,并不用于限制本发明,对于本领域的技术人员来说,本发明实施例可以有各种更改和变化。凡在本发明的精神和原则之内,所作的任何修改、等同替换、改进等,均应包含在本发明的保护范围之内。

Claims (10)

  1. 一种结构光三维扫描***的标定方法,其特征在于,包括:
    根据初始标定参数,对标定平面扫描,得到标定平面的三维点云数据;所述标定平面上设置有多个均匀分布的标记点;
    根据第一指标、第二指标和第三指标与标定参数的关系,找到结构光三维扫描***的最优标定参数;所述第一指标为:根据所述标记点对应的三维点云数据,计算得到的每相临两个标记点之间的距离与所述两个标记点之间的实际距离差异的平均值;所述第二指标为:根据所述标记点对应的三维点云数据,计算得到的每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度与所述两条直线之间的实际夹角角度差异的平均值;所述第三指标为:根据标定平面的三维点云数据,计算得到的标定平面的平面度与所述标定平面的实际平面度的差异。
  2. 如权利要求1所述的结构光三维扫描***的标定方法,其特征在于,根据第一指标、第二指标和第三指标与标定参数的关系,找到结构光三维扫描***的最优标定参数,包括:
    建立所述第一指标、第二指标和第三指标与标定参数的全局优化函数;
    以所述初始标定参数为所述全局优化函数的初始值,迭代找到最优标定参数,对于每个迭代周期均执行以下操作:
    根据所述标记点对应的三维点云数据,计算所述第一指标和第二指标;根据标定平面的三维点云数据,计算所述第三指标;直到找到第一指标、第二指标和第三指标中一个或多个最小时对应的标定参数,作为结构光三维扫描***的最优标定参数。
  3. 如权利要求2所述的结构光三维扫描***的标定方法,其特征在于,以所述初始标定参数为所述全局优化函数的初始值,迭代找到最优标定参数,包括:
    以所述初始标定参数为全局优化函数的初始值,设定标定参数的范围阈值,迭代找到最优标定参数。
  4. 如权利要求1至3任一权利要求所述的结构光三维扫描***的标定方法,其特征在于,所述标记点的形状为圆形。
  5. 如权利要求1至3任一权利要求所述的结构光三维扫描***的标定方法,其特征在于,所述标定平面包括:平面物体以及粘贴在平面物体上的、满足结构光三维扫描***扫描范围的白纸;所述白纸上设置有多个均匀分布的标记点。
  6. 如权利要求1至3任一权利要求所述的结构光三维扫描***的标定方法,其特征在于,所述标定参数包括:结构光三维扫描***中相机和摄影机的内部参数和外部参数。
  7. 如权利要求1至3任一权利要求所述的结构光三维扫描***的标定方法,其特征在于,所述第三指标具体为:根据从标定平面上均匀采样的三维点云数据,计算得到的标定平面的平面度与所述标定平面的实际平面度的差异。
  8. 如权利要求2至3任一权利要求所述的结构光三维扫描***的标定方法,其特征在于,所述全局优化函数为:
    min Fobj(parasSLS)
    s.t.;
    Figure PCTCN2015098936-appb-100001
    其中,Fobj为全局优化函数的目标函数:Fobj=∑(Ed+α·Ep+β·Ea);α,β满足:Ed≈α·Ep≈β·Ea
    Figure PCTCN2015098936-appb-100002
    是标定参数的上限,
    Figure PCTCN2015098936-appb-100003
    是标定参数的下限;Ed为第一指标:每相临两个标记点之间的距离与所述两个标记点之间的实际距离差异的平均值;Ea为第二指标:根据所述标记点对应的三维点云数据,计算得到的每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度与所述两条直线之间的实际夹角角度差异的平均值;Ep为第三指标:根据标定平面的三维点云数据,计算得到的标定平面的平面度与所述标定平面的实际平面度的差异;α和β是经验参数,用来调整Ed、Ea和Ep的数量级。
  9. 一种结构光三维扫描***的标定装置,其特征在于,包括:
    三维点云数据获取模块,用于根据初始标定参数,对标定平面扫描,得到标定平面的三维点云数据;所述标定平面上设置有多个均匀分布的标记点;
    最优标定参数计算模块,用于根据第一指标、第二指标和第三指标与标定参数的关系,找到结构光三维扫描***的最优标定参数;所述第一指标为:根据所述标记点对应的三维点云数据,计算得到的每相临两个标记点之间的距离与所述两个标记点之间的实际距离差异的平均值;所述第二指标为:根据所述标记点对应的三维点云数据,计算得到的每个标记点和它相临的两个标记点连接形成的两条直线之间的夹角角度与所述两条直线之间的实际夹角角度差异的平均值;所述第三指标为:根据标定平面的三维点云数据,计算得到的标定平面的平面度与所述标定平面的实际平面度的差异。
  10. 如权利要求9所述的结构光三维扫描***的标定装置,其特征在于,所述最优标定参数计算模块具体用于:
    建立所述第一指标、第二指标和第三指标与标定参数的全局优化函数;
    以所述初始标定参数为所述全局优化函数的初始值,迭代找到最优标定参数,对于每个迭代周期均执行以下操作:
    根据所述标记点对应的三维点云数据,计算所述第一指标和第二指标;根据标定平面的三维点云数据,计算所述第三指标;直到找到第一指标、第二指标和第三指标中一个或多个最小时对应的标定参数,作为结构光三维扫描***的最优标定参数。
PCT/CN2015/098936 2015-12-14 2015-12-25 结构光三维扫描***的标定方法及装置 WO2017101150A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510925237.5A CN105551039B (zh) 2015-12-14 2015-12-14 结构光三维扫描***的标定方法及装置
CN201510925237.5 2015-12-14

Publications (1)

Publication Number Publication Date
WO2017101150A1 true WO2017101150A1 (zh) 2017-06-22

Family

ID=55830215

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/098936 WO2017101150A1 (zh) 2015-12-14 2015-12-25 结构光三维扫描***的标定方法及装置

Country Status (2)

Country Link
CN (1) CN105551039B (zh)
WO (1) WO2017101150A1 (zh)

Cited By (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108089198A (zh) * 2017-12-11 2018-05-29 同方威视技术股份有限公司 三维扫描装置、机器人及数据处理方法
WO2019039996A1 (en) * 2017-08-25 2019-02-28 Maker Trading Pte Ltd ARTIFICIAL VISION SYSTEM AND METHOD FOR IDENTIFYING LOCATIONS OF TARGET ELEMENTS
CN109920003A (zh) * 2017-12-12 2019-06-21 广东虚拟现实科技有限公司 相机标定检测方法、装置以及设备
CN110440708A (zh) * 2018-05-04 2019-11-12 苏州玻色智能科技有限公司 一种用于三维白光扫描设备的标准件及其标定方法
CN110598239A (zh) * 2019-07-19 2019-12-20 中建安装集团有限公司 基于轨行区点云大数据的应用方法
CN111028280A (zh) * 2019-12-09 2020-04-17 西安交通大学 井字结构光相机***及进行目标有尺度三维重建的方法
CN111445534A (zh) * 2020-04-03 2020-07-24 北京四维智联科技有限公司 一种用于全自动泊车的相机高精度自动标定方法及装置
CN111627113A (zh) * 2020-05-25 2020-09-04 易视智瞳科技(深圳)有限公司 3d线扫建模水平角度的标定方法、装置、设备及存储介质
CN111735487A (zh) * 2020-05-18 2020-10-02 清华大学深圳国际研究生院 传感器、传感器标定方法与设备、存储介质
CN112270746A (zh) * 2020-11-06 2021-01-26 太原科技大学 基于邻域协方差特征参数阈值的铝合金3d打印点云精简算法
CN112581605A (zh) * 2020-12-24 2021-03-30 西安中科光电精密工程有限公司 一种结构光三维重建校正方法及装置
CN112750165A (zh) * 2019-10-29 2021-05-04 商汤集团有限公司 参数标定方法、智能驾驶方法及其装置、设备和存储介质
WO2021115331A1 (zh) * 2019-12-13 2021-06-17 深圳市瑞立视多媒体科技有限公司 基于三角测量的坐标定位方法、装置、设备及存储介质
CN113034616A (zh) * 2021-03-31 2021-06-25 黑芝麻智能科技(上海)有限公司 车辆环视***的相机外参标定方法、***及环视***
CN113133319A (zh) * 2019-10-31 2021-07-16 深圳市大疆创新科技有限公司 标定板、测试角分辨率的方法、装置及计算机存储介质
CN113487680A (zh) * 2021-06-29 2021-10-08 湖南泽塔科技有限公司 一种单幅图像内外参数标定方法及***
US11549805B2 (en) 2019-12-09 2023-01-10 Industrial Technology Research Institute Projecting apparatus and projecting calibration method
CN115937069A (zh) * 2022-03-24 2023-04-07 北京小米移动软件有限公司 零件检测方法、装置、电子设备及存储介质
CN110849389B (zh) * 2019-10-24 2023-08-29 西北大学 一种基于空间球的两个三维点云采集***外参数标定方法

Families Citing this family (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106918302B (zh) * 2017-03-13 2019-09-27 东华大学 一种三维扫描仪自适应标定方法
CN107133989B (zh) * 2017-06-12 2020-11-06 中国科学院长春光学精密机械与物理研究所 一种三维扫描***参数标定方法
CN108921902B (zh) * 2018-06-06 2020-05-08 贝壳技术有限公司 一种修正结构光相机偏差的方法及装置
CN109061609B (zh) * 2018-08-01 2021-09-07 歌尔光学科技有限公司 Tof模组标定装置及方法
JP7139953B2 (ja) * 2019-01-08 2022-09-21 オムロン株式会社 三次元形状測定装置、三次元形状測定方法及びプログラム
CN109977466B (zh) * 2019-02-20 2021-02-02 深圳大学 一种三维扫描视点规划方法、装置及计算机可读存储介质
CN111833392A (zh) * 2019-04-16 2020-10-27 杭州思看科技有限公司 标记点多角度扫描方法、***及装置
CN110207622B (zh) * 2019-06-30 2021-04-20 武汉尺子科技有限公司 一种三维扫描仪标定方法及***
CN111161339B (zh) * 2019-11-18 2020-11-27 珠海随变科技有限公司 一种间距测量方法、装置、设备以及计算机可读介质
US20230228560A1 (en) * 2019-11-19 2023-07-20 Hewlett-Packard Development Company, L.P. Determining a preferred region of a scanner
CN110930382A (zh) * 2019-11-19 2020-03-27 广东博智林机器人有限公司 基于标定板特征点提取的点云拼接精度评估方法及***
GB2594111B (en) 2019-12-18 2023-06-07 Motional Ad Llc Camera-to-LiDAR calibration and validation
CN111486802B (zh) * 2020-04-07 2021-04-06 东南大学 基于自适应距离加权的旋转轴标定方法
CN112082512B (zh) * 2020-09-08 2023-04-14 深圳广成创新技术有限公司 一种相位测量偏折术的标定优化方法、装置及计算机设备
WO2023277889A1 (en) * 2021-06-29 2023-01-05 Hewlett-Packard Development Company, L.P. Calibration of 3d scanning systems
CN113970297A (zh) * 2021-11-19 2022-01-25 江苏智库智能科技有限公司 堆垛机货位列数据的标定测量方法
CN114037768A (zh) * 2021-12-29 2022-02-11 天远三维(天津)科技有限公司 一种多套跟踪式扫描仪联合标定的方法和装置
CN116772746A (zh) * 2023-08-17 2023-09-19 湖南视比特机器人有限公司 利用射灯图案检测进行平面度轮廓度测量方法及存储介质

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070115484A1 (en) * 2005-10-24 2007-05-24 Peisen Huang 3d shape measurement system and method including fast three-step phase shifting, error compensation and calibration
CN101231750A (zh) * 2008-02-21 2008-07-30 南京航空航天大学 一种双目立体测量***的标定方法
CN101245994A (zh) * 2008-03-17 2008-08-20 南京航空航天大学 物体表面三维轮廓结构光测量***的标定方法
CN101419061A (zh) * 2008-12-08 2009-04-29 北京航空航天大学 一种镜像式结构光视觉测量***和测量方法
CN101697233A (zh) * 2009-10-16 2010-04-21 长春理工大学 一种基于结构光的三维物体表面重建方法
CN101763643A (zh) * 2010-01-07 2010-06-30 浙江大学 一种结构光三维扫描仪***自动标定方法
US20140078260A1 (en) * 2012-09-20 2014-03-20 Brown University Method for generating an array of 3-d points
CN103942796A (zh) * 2014-04-23 2014-07-23 清华大学 一种高精度的投影仪-摄像机标定***及标定方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101673399A (zh) * 2009-09-29 2010-03-17 浙江工业大学 编码结构光三维视觉***的标定方法
CN103985157A (zh) * 2014-05-30 2014-08-13 深圳先进技术研究院 一种结构光三维扫描方法及***

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070115484A1 (en) * 2005-10-24 2007-05-24 Peisen Huang 3d shape measurement system and method including fast three-step phase shifting, error compensation and calibration
CN101231750A (zh) * 2008-02-21 2008-07-30 南京航空航天大学 一种双目立体测量***的标定方法
CN101245994A (zh) * 2008-03-17 2008-08-20 南京航空航天大学 物体表面三维轮廓结构光测量***的标定方法
CN101419061A (zh) * 2008-12-08 2009-04-29 北京航空航天大学 一种镜像式结构光视觉测量***和测量方法
CN101697233A (zh) * 2009-10-16 2010-04-21 长春理工大学 一种基于结构光的三维物体表面重建方法
CN101763643A (zh) * 2010-01-07 2010-06-30 浙江大学 一种结构光三维扫描仪***自动标定方法
US20140078260A1 (en) * 2012-09-20 2014-03-20 Brown University Method for generating an array of 3-d points
CN103942796A (zh) * 2014-04-23 2014-07-23 清华大学 一种高精度的投影仪-摄像机标定***及标定方法

Cited By (32)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2019039996A1 (en) * 2017-08-25 2019-02-28 Maker Trading Pte Ltd ARTIFICIAL VISION SYSTEM AND METHOD FOR IDENTIFYING LOCATIONS OF TARGET ELEMENTS
US11080880B2 (en) 2017-08-25 2021-08-03 Maker Trading Pte Ltd Machine vision system and method for identifying locations of target elements
CN108089198A (zh) * 2017-12-11 2018-05-29 同方威视技术股份有限公司 三维扫描装置、机器人及数据处理方法
CN108089198B (zh) * 2017-12-11 2023-12-19 同方威视技术股份有限公司 三维扫描装置、机器人及数据处理方法
CN109920003A (zh) * 2017-12-12 2019-06-21 广东虚拟现实科技有限公司 相机标定检测方法、装置以及设备
CN109920003B (zh) * 2017-12-12 2023-09-15 广东虚拟现实科技有限公司 相机标定检测方法、装置以及设备
CN110440708A (zh) * 2018-05-04 2019-11-12 苏州玻色智能科技有限公司 一种用于三维白光扫描设备的标准件及其标定方法
CN110440708B (zh) * 2018-05-04 2024-06-07 苏州玻色智能科技有限公司 一种用于三维白光扫描设备的标准件及其标定方法
CN110598239B (zh) * 2019-07-19 2023-11-17 中国建设基础设施有限公司 基于轨行区点云大数据的应用方法
CN110598239A (zh) * 2019-07-19 2019-12-20 中建安装集团有限公司 基于轨行区点云大数据的应用方法
CN110849389B (zh) * 2019-10-24 2023-08-29 西北大学 一种基于空间球的两个三维点云采集***外参数标定方法
CN112750165B (zh) * 2019-10-29 2023-12-15 商汤集团有限公司 参数标定方法、智能驾驶方法及其装置、设备和存储介质
CN112750165A (zh) * 2019-10-29 2021-05-04 商汤集团有限公司 参数标定方法、智能驾驶方法及其装置、设备和存储介质
CN113133319A (zh) * 2019-10-31 2021-07-16 深圳市大疆创新科技有限公司 标定板、测试角分辨率的方法、装置及计算机存储介质
CN111028280B (zh) * 2019-12-09 2022-06-03 西安交通大学 井字结构光相机***及进行目标有尺度三维重建的方法
CN111028280A (zh) * 2019-12-09 2020-04-17 西安交通大学 井字结构光相机***及进行目标有尺度三维重建的方法
US11549805B2 (en) 2019-12-09 2023-01-10 Industrial Technology Research Institute Projecting apparatus and projecting calibration method
WO2021115331A1 (zh) * 2019-12-13 2021-06-17 深圳市瑞立视多媒体科技有限公司 基于三角测量的坐标定位方法、装置、设备及存储介质
CN111445534B (zh) * 2020-04-03 2023-04-21 北京四维智联科技有限公司 一种用于全自动泊车的相机高精度自动标定方法及装置
CN111445534A (zh) * 2020-04-03 2020-07-24 北京四维智联科技有限公司 一种用于全自动泊车的相机高精度自动标定方法及装置
CN111735487A (zh) * 2020-05-18 2020-10-02 清华大学深圳国际研究生院 传感器、传感器标定方法与设备、存储介质
CN111735487B (zh) * 2020-05-18 2023-01-10 清华大学深圳国际研究生院 传感器、传感器标定方法与设备、存储介质
CN111627113B (zh) * 2020-05-25 2023-09-19 易视智瞳科技(深圳)有限公司 3d线扫建模水平角度的标定方法、装置、设备及存储介质
CN111627113A (zh) * 2020-05-25 2020-09-04 易视智瞳科技(深圳)有限公司 3d线扫建模水平角度的标定方法、装置、设备及存储介质
CN112270746B (zh) * 2020-11-06 2023-06-16 太原科技大学 基于邻域协方差特征参数阈值的铝合金3d打印点云精简算法
CN112270746A (zh) * 2020-11-06 2021-01-26 太原科技大学 基于邻域协方差特征参数阈值的铝合金3d打印点云精简算法
CN112581605A (zh) * 2020-12-24 2021-03-30 西安中科光电精密工程有限公司 一种结构光三维重建校正方法及装置
CN112581605B (zh) * 2020-12-24 2024-05-14 西安中科光电精密工程有限公司 一种结构光三维重建校正方法及装置
CN113034616A (zh) * 2021-03-31 2021-06-25 黑芝麻智能科技(上海)有限公司 车辆环视***的相机外参标定方法、***及环视***
CN113487680A (zh) * 2021-06-29 2021-10-08 湖南泽塔科技有限公司 一种单幅图像内外参数标定方法及***
CN115937069A (zh) * 2022-03-24 2023-04-07 北京小米移动软件有限公司 零件检测方法、装置、电子设备及存储介质
CN115937069B (zh) * 2022-03-24 2023-09-19 北京小米移动软件有限公司 零件检测方法、装置、电子设备及存储介质

Also Published As

Publication number Publication date
CN105551039B (zh) 2017-12-08
CN105551039A (zh) 2016-05-04

Similar Documents

Publication Publication Date Title
WO2017101150A1 (zh) 结构光三维扫描***的标定方法及装置
CN108151671B (zh) 一种三维数字成像传感器、三维扫描***及其扫描方法
CN102376089B (zh) 一种标靶校正方法及***
CN107133989B (zh) 一种三维扫描***参数标定方法
CN108053450B (zh) 一种基于多约束的高精度双目相机标定方法
CN102364299B (zh) 一种多个结构光投影三维型面测量头的标定技术
CN113409382B (zh) 车辆损伤区域的测量方法和装置
CN107886547B (zh) 一种鱼眼相机标定方法及***
CN108759669B (zh) 一种室内自定位三维扫描方法及***
KR20220080011A (ko) 고정밀 교정 시스템 및 방법
CN109827521B (zh) 一种快速多线结构光视觉测量***标定方法
CN110148174A (zh) 标定板、标定板识别方法及装置
CN109297436A (zh) 双目线激光立体测量基准标定方法
CN109272555B (zh) 一种rgb-d相机的外部参数获得及标定方法
WO2018006246A1 (zh) 四相机组平面阵列特征点匹配方法及基于其的测量方法
CN109307483A (zh) 一种基于结构光***几何约束的相位展开方法
CN102829736A (zh) 一种三维指纹传感***
WO2020199439A1 (zh) 基于单双目混合测量的三维点云计算方法
Ye et al. An accurate 3D point cloud registration approach for the turntable-based 3D scanning system
WO2023201578A1 (zh) 单目激光散斑投影***的外参数标定方法和装置
Fernandez et al. Planar-based camera-projector calibration
Daftry et al. Flexible and User-Centric Camera Calibration using Planar Fiducial Markers.
CN102930551A (zh) 利用圆心的投影坐标和极线求解摄像机内参数
CN107367245B (zh) 光学三维轮廓测量中的无效点探测与剔除方法
CN102999895B (zh) 利用两个同心圆线性求解摄像机内参数

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: 15910619

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 08.10.2018)

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 08/10/2018)

122 Ep: pct application non-entry in european phase

Ref document number: 15910619

Country of ref document: EP

Kind code of ref document: A1