CN114964316B - Position and attitude calibration method and device, and method and system for measuring target to be measured - Google Patents

Position and attitude calibration method and device, and method and system for measuring target to be measured Download PDF

Info

Publication number
CN114964316B
CN114964316B CN202210891478.2A CN202210891478A CN114964316B CN 114964316 B CN114964316 B CN 114964316B CN 202210891478 A CN202210891478 A CN 202210891478A CN 114964316 B CN114964316 B CN 114964316B
Authority
CN
China
Prior art keywords
camera
relative position
sina
cosa
image
Prior art date
Legal status (The legal status 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 status listed.)
Active
Application number
CN202210891478.2A
Other languages
Chinese (zh)
Other versions
CN114964316A (en
Inventor
曹动
饶旭
肖永恒
王宇轩
刘小舟
张建南
何江
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Rocketech Technology Corp ltd
Original Assignee
Rocketech Technology Corp ltd
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 Rocketech Technology Corp ltd filed Critical Rocketech Technology Corp ltd
Priority to CN202210891478.2A priority Critical patent/CN114964316B/en
Publication of CN114964316A publication Critical patent/CN114964316A/en
Application granted granted Critical
Publication of CN114964316B publication Critical patent/CN114964316B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C25/00—Manufacturing, calibrating, cleaning, or repairing instruments or devices referred to in the other groups of this subclass
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B11/00—Measuring arrangements characterised by the use of optical techniques
    • G01B11/002—Measuring arrangements characterised by the use of optical techniques for measuring two or more coordinates
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B21/00—Measuring arrangements or details thereof, where the measuring technique is not covered by the other groups of this subclass, unspecified or not relevant
    • G01B21/02—Measuring arrangements or details thereof, where the measuring technique is not covered by the other groups of this subclass, unspecified or not relevant for measuring length, width, or thickness
    • G01B21/04—Measuring arrangements or details thereof, where the measuring technique is not covered by the other groups of this subclass, unspecified or not relevant for measuring length, width, or thickness by measuring coordinates of points
    • G01B21/042—Calibration or calibration artifacts
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/20—Image signal generators
    • H04N13/204—Image signal generators using stereoscopic image cameras
    • H04N13/246—Calibration of cameras

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Manufacturing & Machinery (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Length Measuring Devices By Optical Means (AREA)

Abstract

The invention relates to the technical field of vision measurement and image detection, and provides a position and posture calibration method and device, and a method and system for measuring a target to be measured. The calibration method comprises the following steps: the stereoscopic vision measurement system comprises a first camera and a second camera, n bright spots are projected to a target to be measured, the stereoscopic vision camera shoots a target image to be measured with the n bright spots, image points corresponding to the n bright spots in the target image to be measured are extracted, and image points of the second camera and image points of the first camera are matched with each other; calculating the normalized image coordinates of the image points of the stereoscopic vision camera; and respectively solving the relative position posture of each second camera and the first camera according to the image points with the same name and the normalized image coordinates of the image points of the first camera and the second camera. The scheme can ensure the accuracy of the measured relative position and attitude and is not limited to various application occasions.

Description

Position and attitude calibration method and device, and method and system for measuring target to be measured
Technical Field
The invention relates to the technical field of vision measurement and image detection, in particular to a method and a device for calibrating the relative position and attitude of a stereoscopic vision measurement system, a method for measuring a target to be measured and a self-calibrated measurement system.
Background
The vision measurement is an advanced system which takes computer vision as a theory, adopts an advanced image sensor with high density, low noise and small distortion, and effectively processes binary or gray level images through a high-speed real-time image acquisition system, a special image hardware processing system and a high-performance computer. Usually, a plurality of cameras are used to shoot the target to be measured and then measure the target to be measured.
The accuracy of the relative position and orientation between the cameras of a stereo vision measurement system is critical to the accuracy of the stereo vision measurement results. Small relative attitude errors between the cameras can cause large stereo vision measurement errors.
In order to reduce the error of the relative position and attitude, the prior art generally has the following steps: firstly, accurately calibrating the relative position and posture of each camera in advance, and ensuring that the system is stable and the parameters are unchanged when the system is used; in this way, when the use environment conditions are severe (such as impact vibration, etc.), it is difficult to ensure the stability of the measurement system structure and the stability and invariability of the camera parameters. Secondly, each camera is fixedly connected with an attitude sensor, and the pose variation of the device is read in real time in the using process so as to compensate; by adopting the mode, the sensors fixedly connected are additionally arranged, the high-precision pose sensor is high in cost, and the attitude sensor represented by inertial navigation generally has the problems of time accumulation error and low long-time application precision. Thirdly, control point targets with accurately known coordinates are distributed in a measurement view field, and real-time parameters of each camera are calibrated during measurement; and (3) the condition that the control point is not arranged in the measurement visual field is possible, such as the condition that the target to be measured in the small measurement visual field occupies the full visual field, or the condition that the measurement visual field is continuously changed when the measurement platform is moved. Therefore, the prior art has limited application occasions.
Therefore, it is necessary to develop a method and an apparatus for calibrating the relative position and orientation of a stereoscopic vision measurement system, a method for measuring an object to be measured, and a self-calibrated measurement system, which are not limited to various applications while ensuring the accuracy of the measurement of the relative position and orientation.
Disclosure of Invention
The invention aims to provide a method and a device for calibrating the relative position and the attitude of a stereoscopic vision measuring system, a method for measuring a target to be measured and a self-calibrated measuring system, which can ensure the accuracy of the measured relative position and the attitude and are not limited to various application occasions.
In order to solve the above technical problem, as an aspect of the present invention, there is provided a method for calibrating a relative position and orientation of a stereoscopic vision measurement system, including the steps of:
s1: the stereovision measurement system is used for shooing the target that awaits measuring, and the stereovision measurement system includes m stereovision cameras, and m is more than or equal to 2's natural number, and m stereovision cameras include: a first camera and m-1 second cameras; shooting a target to be detected from different angles by m stereoscopic vision cameras; projecting n bright spots to a target to be detected, wherein the n bright spots are all in the shooting range of a stereoscopic vision camera; n is a natural number greater than or equal to 3;
s2: all stereoscopic vision cameras shoot a target image to be detected with n bright spots, image points corresponding to the n bright spots in the target image to be detected are extracted, and image points of a second camera are matched with image points of a first camera in a same name mode;
s3: calculating the normalized image coordinates of the image points of the stereoscopic vision camera;
s4: and respectively solving the relative position posture of each second camera and the first camera according to the image points with the same name and the normalized image coordinates of the image points of the first camera and the second camera.
According to an exemplary embodiment of the present invention, in step S3, the method for calculating the normalized image coordinates of the image point of the stereoscopic vision camera includes:
calculating the normalized image coordinates of the image points according to the internal parameters of the stereoscopic vision camera;
the internal parameters of the stereoscopic vision camera comprise principal point coordinates and equivalent focal length;
the image point normalized image coordinates of the first camera are:
Figure 54587DEST_PATH_IMAGE001
；
the image point normalized image coordinates of the second camera are:
Figure 156273DEST_PATH_IMAGE002
；
wherein the principal point coordinate of the first camera is (C)x0, Cy0) The equivalent focal length of the first camera is (Fx)0, Fy0) The principal point coordinate of the second camera is (C)xj, Cyj) The equivalent focal length of the second camera is (F)xj, Fyj)，j=1,2，...，m-1；i=0,1，...，n-1。
According to an exemplary embodiment of the present invention, in step S4, the method for separately solving the relative position and orientation between each second camera and the first camera according to the image point with the same name and the normalized image coordinates of the image points of the first camera and the second camera includes:
setting an initial value of the relative position posture of the camera;
establishing a relative position attitude equation according to the relation between the first camera and the second camera;
listing a relative position attitude equation according to all the image points with the same name;
and solving a relative attitude equation according to the initial value and the listed relative position attitude equation to obtain a corrected value.
According to an exemplary embodiment of the present invention, the method for separately solving the relative position and orientation of each second camera and the first camera according to the image point with the same name and the normalized image coordinates of the image points of the first camera and the second camera further includes: after the correction value is obtained, the correction value is used as a new initial value to continuously solve a new correction value; and repeating the iteration initial value and solving a new correction value until the iteration initial value is repeated until the specified times and/or the error of the initial value calculation of the two times of updating is less than the specified threshold value or the measurement of the target to be measured is finished.
According to an example embodiment of the present invention, the relative attitude equation is:
Figure 284504DEST_PATH_IMAGE003
；
the second camera and the first camera relative position pose comprises a translation vector and/or a rotation angle;
wherein the translation vector is
Figure 875977DEST_PATH_IMAGE004
The rotation matrix is
Figure 773264DEST_PATH_IMAGE005
At a rotation angle of
Figure 311430DEST_PATH_IMAGE006
The rotation matrix is a trigonometric combination of rotation angles; x, y, z are mutually perpendicular axes, txj、tyj、tzjRepresenting the length of movement of the camera in three axes, azj、ayj、axjThe angle of the camera rotating around the three axes of z, y and x in turn, r0j=cosayj×cosazj、r1j=sinaxj×sinayj×cosazj-cosaxj×sinazj、r2j=cosaxj×sinayj×cosazj+sinaxj×sinaxzj、r3j=cosayj×sinazj、r4j=sinaxj×sinayj×sinazj+cosaxj×cosazj、r5j=cosaxj×sinayj×sinazj-sinaxj×cosazj、r6j=-sinayj、r7j=sinaxj×cosayj、r8j=cosaxj×cosayj。
According to an exemplary embodiment of the present invention, the method for solving the relative attitude equation according to the initial value and the listed relative position attitude equation comprises:
adopting a nonlinear optimization method to obtain the final product at the initial value
Figure 458116DEST_PATH_IMAGE007
Calculating a deviation of the translation vector and/or the rotation angle;
to pair
Figure 779244DEST_PATH_IMAGE008
Performing Taylor expansion to obtain a linear equation set of correction quantity related to the translation vector and/or the rotation angle;
solving the linear equation set by a least square method to obtain a correction quantity;
and correcting the translation vector and/or the rotation angle according to the correction amount to obtain a correction value.
The nonlinear optimization method includes a newton iteration method.
As a second aspect of the present invention, a method for measuring a target to be measured is provided, wherein a stereo vision camera of a vision measuring system is calibrated by using the calibration method for the relative position and posture of the stereo vision measuring system;
and measuring the object to be measured by using a stereoscopic vision camera.
As a third aspect of the present invention, there is provided a relative position posture calibration apparatus for a stereo vision measurement system, comprising:
a scattered bright spot irradiator capable of irradiating n bright spots to a target to be measured, wherein n is a natural number greater than or equal to 3;
the homonymous image point matching module is used for extracting image points corresponding to the n bright points in the target image to be detected and respectively matching the image points of the second camera with the image points of the first camera;
the image coordinate calculation module is used for calculating the normalized image coordinates of the image points of the stereoscopic vision camera;
the relative position posture calculation module is used for respectively solving the relative position posture of each second camera and the first camera according to the image points with the same name and the normalized image coordinates of the image points of the first camera and the second camera;
wherein, the stereovision measurement system is used for shooing the target that awaits measuring, and the stereovision measurement system includes m stereovision cameras, and m is more than or equal to 2's natural number, and m stereovision cameras include: a first camera and m-1 second cameras; the m stereoscopic vision cameras shoot the target to be measured from different angles.
According to an exemplary embodiment of the present invention, the scattered bright spot irradiator is a laser.
As a fourth aspect of the present invention, there is provided a self-calibrating measurement system comprising:
the device for calibrating the relative position and posture of the stereoscopic vision measuring system and the stereoscopic vision measuring system are provided.
According to an example embodiment of the present invention, the stereo vision measurement system calibrates the first camera and the second camera according to the relative position pose.
The invention has the beneficial effects that:
the scheme adopts a mode of illuminating bright spots on the target to be detected to calibrate the stereoscopic vision camera, can adapt to the conditions that the surface of the target to be detected has no stable texture and the environmental illumination condition is poor, has simple equipment for spreading the bright spots, has no structural stability and precision requirements in the processes of preparation, installation and use, and has wide application range; the method can calibrate the attitude parameters of the relative position of the stereo vision camera in real time, ensures the precision of stereo vision measurement under the conditions of parameter disturbance, change and the like of a stereo vision measurement system, and adapts to the conditions that the motion of a measurement scene changes, the region of the target to be measured cannot be provided with a calibrated target point and the like because bright spots are arranged on the target to be measured and a mode of actively irradiating the region of the target to be measured along with the stereo vision measurement system to obtain characteristic points is adopted.
Drawings
Fig. 1 schematically shows a configuration of a relative position and orientation calibration device of a stereo vision measurement system.
Fig. 2 schematically shows a positional relationship diagram when the relative position and orientation of the stereo vision measurement system are calibrated.
Fig. 3 schematically shows an imaging diagram of a stereoscopic vision camera.
Wherein 1-the target to be measured, 2-the scattered light spot irradiator, C0A first camera, Cj-a second camera.
Detailed Description
The following detailed description of embodiments of the invention, but the invention can be practiced in many different ways, as defined and covered by the claims.
As a first embodiment of the present invention, there is provided a relative position and orientation calibration device for a stereo vision measurement system, as shown in fig. 1, including: the device comprises a distributed bright spot irradiator, a homonymy image point matching module, an image coordinate calculation module and a relative position and attitude calculation module.
The scattered bright spot irradiator may irradiate n bright spot irradiators to the target to be measured, n being a natural number greater than or equal to 3. The scattered light spot irradiator uses a laser.
Homonymous image point matching module and first camera C0And a second camera CjAnd the connection is used for extracting image points corresponding to the n bright points in the target image to be detected, and respectively matching the image points of the second camera with the image points of the first camera by means of the same name.
Image coordinate calculation module and first camera C0Second camera CjAnd the connection is used for calculating the normalized image coordinates of the image points of the stereoscopic vision camera.
The relative position and posture calculation module is connected with the homonymous image point matching module and the image coordinate calculation module and is used for respectively solving the relative position and posture of each second camera and each first camera according to the homonymous image points and the normalized image coordinates of the image points of the first camera and the second camera.
Wherein, stereovision measurement system is used for shooing the target that awaits measuring, and stereovision measurement system includes m stereovision cameras, and m is the natural number more than or equal to 2, and m stereovision cameras include: a first camera and m-1 second cameras; the m stereoscopic vision cameras shoot the target to be measured from different angles.
The device of this scheme of adoption can solve the relative position gesture of each second camera and first camera in real time, and the accuracy is high.
As a second embodiment of the present invention, there is provided a method for calibrating a relative position and orientation of a stereo vision measurement system, which is implemented by the apparatus of the first embodiment. As shown in fig. 2, the stereoscopic vision measuring system is used for shooting an object 1 to be measured, and includes m stereoscopic vision cameras, where m is a natural number greater than or equal to 2, and the m stereoscopic vision cameras include: a first camera C0And m-1 second cameras CjJ =1, 2.., m-1. The m stereoscopic vision cameras shoot the object 1 to be measured from different angles. M in FIG. 2 is 2, with a first camera C0For reference, the second camera C is calculated separatelyjAnd a first camera C0In a relative position posture of the camera, adding a second camera CjCalculates one more second camera CjAnd a first camera C0The relative position and posture of the invention do not influence the scheme to be protected. Second camera CjIndicating the jth second camera.
Before calibration, the internal parameters of the stereo vision camera have been calibrated since they are unchanged. The internal parameters of the stereoscopic vision camera include principal point coordinates and an equivalent focal length. As shown in FIG. 3, the true geographic location coordinate of the stereovision camera's optical center is (X)0，Y0，Z0) The coordinates of the optical center, i.e., the principal point, imaged by the camera are (x)c，yc). The image surface of the camera image has a pixel size of (d)x，dy) Focal length of camera is f, equivalent focal length f in x directionx=f/dxEquivalent focal length f in the y directiony=f/dyThe equivalent focal length is the normalized focal length on the x-axis and the y-axis. In the figureIn 3, the coordinates of the true geographic position of the bright point are (X, Y, Z), the optical center and the bright point are connected into a straight line, and an image point (X, Y) is displayed on the image plane of the camera.
The calibration method comprises the following steps:
s1: the scattered bright spot irradiator 2 projects n bright spots to the target 1 to be measured, wherein the n bright spots are all in the shooting range (measurement field) of the stereoscopic vision camera; n is a natural number of 3 or more. The number of n is determined according to the number of parameters of the relative position posture which needs to be solved finally, and the equation can be solved only if the number of n is more than or equal to the number of parameters of the relative position posture. The n bright spots are distributed uniformly as much as possible.
S2: all stereoscopic vision cameras shoot images of a target 1 to be detected with n bright spots, the image point matching module with the same name extracts image points corresponding to the n bright spots in the image of the target 1 to be detected, and the second cameras C are respectively usedjImage point and the first camera C0The image points of (2) are matched with the image points of the same name.
The scattered bright spot irradiator 2 irradiates n bright spots, which are respectively a bright spot 0, a bright spot 1, a bright spot 2, a bright spot, and a bright spot n-1. The image points corresponding to the n bright points in the image of the target 1 to be detected are an image point 0, an image point 1, an image point 2, an image point n-1. A second camera CjImage point 0 and the first camera C0Is matched with the second camera CjImage point 1 and the first camera C0Match the image point 1 of the second camera C, and so onjImage point n-1 and the first camera C0Is matched to the image point n-1. First camera C0Image point i and the second camera CjThe image point i of (a) is the image point of the same name. First camera C0Has the coordinate of (x)0,i, y0,i) Second camera CjHas the coordinate of (x)j,i, yj,i)，i=0,1，...，n-1。
S3: and calculating the normalized image coordinates of the image points of the stereoscopic vision camera.
And calculating the normalized image coordinates of the image points according to the internal parameters of the stereoscopic vision camera.
First camera C0Image point i normalization imageThe coordinates are:
Figure 593354DEST_PATH_IMAGE001
；
second camera CjThe image point i normalized image coordinates are:
Figure 517403DEST_PATH_IMAGE002
；
wherein the first camera C0Has a principal point coordinate of (C)x0, Cy0) First camera C0Has an equivalent focal length of (F)x0, Fy0) Second camera CjHas a principal point coordinate of (C)xj, Cyj) Second camera CjHas an equivalent focal length of (F)xj, Fyj)，j=1,2，...，m-1；i=0,1，...，n-1。
The coordinates of the image point on the image plane can be obtained as the x and y coordinates, as shown in fig. 3, the image point is on the connecting line of the bright point and the optical center, and as long as the bright point is on this line, the image point at the same position is present on the image, so the z coordinate of the image point is unknown, and it is necessary to set the z of the image point to 1, and normalize the coordinates of the image point. The coordinate values of the normalized image represent the values after x, y, z normalization from top to bottom.
S4: according to the same name image point and the first camera C0And a second camera CjRespectively solving the normalized image coordinates of the image points to obtain each second camera CjAnd a first camera C0Relative position posture.
Each second camera CjAnd a first camera C0The calculation method of the relative position and posture is the same, and a second camera C is usedjAnd a first camera C0Calculation of the relative position posture is an example.
Second camera Cj and first camera C0The relative position pose includes a translation vector and/or a rotation angle.
The translation vector is
Figure 885805DEST_PATH_IMAGE004
The rotation matrix is
Figure 807363DEST_PATH_IMAGE005
At a rotation angle of
Figure 148083DEST_PATH_IMAGE006
The rotation matrix is a trigonometric combination of rotation angles. Each element of the rotation matrix is a second camera CjRelative to the first camera C0Angle of rotation axj，ayj，azjThe trigonometric function combination of (1); x, y, z are mutually perpendicular axes, txj、tyj、tzjRepresenting the length of movement of the camera in three axes, azj、ayj、axjThe angle of the camera rotating around the three axes of z, y and x in turn, r0j=cosayj×cosazj、r1j=sinaxj×sinayj×cosazj-cosaxj×sinazj、r2j=cosaxj×sinayj×cosazj+sinaxj×sinaxzj、r3j=cosayj×sinazj、r4j=sinaxj×sinayj×sinazj+cosaxj×cosazj、r5j=cosaxj×sinayj×sinazj-sinaxj×cosazj、r6j=-sinayj、r7j=sinaxj×cosayj、r8j=cosaxj×cosayj。
The translation vector is three parameters, txj，tyj，tzj(ii) a The rotation angle is three parameters, respectively axj，ayj，azj. If the translation vector is known, only three parameters of the rotation angle need to be solved, and the number of the bright spots n is more than or equal to 3; if the rotation angle is known, only three parameters of the translation vector need to be solved, the number of the bright points n is more than or equal to 3, and the situation that the rotation angle is known is rare because high-precision attitude data is difficult to obtain. If both the translation vector and the rotation angle are unknown, then six solutions need to be solvedThe number of the bright spots n is more than or equal to 6. The larger the number of the bright points n, the more the equation system is solved, the more accurate the obtained numerical value is, but the larger the calculation amount is, generally, the bright points are required to be uniformly distributed in the field of view, and the number of the bright points is dozens or hundreds. Therefore, the number of the bright spots n is preferably 20 to 999.
S401: and setting an initial value of the relative position posture of the camera. The initial value of the position may be set to 0 or to data close to the actual relative position value. The initial value of the pose is a 3 x 3 identity matrix.
S402: according to the first camera C0And a second camera CjThe relationship of (a) establishes a relative position attitude equation.
According to the stereoscopic vision imaging relationship:
Figure 293631DEST_PATH_IMAGE009
；
the formula is related to txj，tyj，tzj，axj，ayj，azjIs converted into a relative attitude equation.
Jth second camera CjAnd a first camera C0The relative attitude equation of (a) is:
Figure 833152DEST_PATH_IMAGE003
。
s403: and listing the relative position and attitude equation according to all the image points with the same name.
Listing a relative position posture equation of the image point 0 with the same name, listing a relative position posture equation of the image point 1 with the same name, and so on, listing a relative position posture equation of the image point n-1 with the same name.
S404: and solving a relative attitude equation according to the initial value and the listed relative position attitude equation to obtain a corrected value.
The method for solving the relative attitude equation according to the initial value and the listed relative position attitude equation comprises the following steps:
adopting a nonlinear optimization method to carry out the operation at an initial value
Figure 823980DEST_PATH_IMAGE010
Calculating a deviation of the translation vector and/or the rotation angle;
to pair
Figure 222469DEST_PATH_IMAGE011
Performing Taylor expansion to obtain a linear equation set of correction quantity related to the translation vector and/or the rotation angle;
solving the linear equation set by a least square method to obtain a correction quantity;
and correcting the translation vector and/or the rotation angle according to the correction amount to obtain a correction value.
The nonlinear optimization method comprises a gradient descent method, a Newton iteration method, a Gaussian Newton method, a Levenberg-Marquardt method, a Dog-leg method and the like. Newton's iterative method is preferred.
The gradient descent method adopts a method of performing first-order taylor approximation on an objective function.
The Newton iteration method needs to calculate a Hessain matrix for the second-order Taylor approximation of the objective function.
Gauss Newton method using JTJ approximates the Hessain matrix.
Levenberg-Marquardt method combines gradient descent method and Gauss Newton method, in order to ensure approximate Hessain matrix positive, at JTAnd adding a damping term to J.
There are specifically the following three cases:
in the first case, if the translation vector and the rotation angle are corrected, txj，tyj，tzj，axj，ayj，azjAre unknown:
adopting a nonlinear optimization method to carry out the operation at an initial value
Figure 804498DEST_PATH_IMAGE012
For translation vector txj，tyj，tzjAnd a rotation angle axj，ayj，azjCalculating a partial derivative;
to pair
Figure 819596DEST_PATH_IMAGE013
Taylor expansion is carried out to obtain a vector t related to translationxj，tyj，tzjAnd a rotation angle axj，ayj，azjCorrection amount dt ofxj，dtyj，dtzj，daxj，dayj，dazjA system of linear equations of (c);
the linear equation set is solved by a least square method to obtain a correction dtxj，dtyj，dtzj，daxj，dayj，dazj；
According to the correction dtxj，dtyj，dtzj，daxj，dayj，dazjFor translation vector txj，tyj，tzjAnd a rotation angle axj，ayj，azjAnd (3) correcting:
Figure 879694DEST_PATH_IMAGE014
and obtaining a correction value.
Second case, if only the translation vector is corrected, txj，tyj，tzjIs unknown, axj，ayj，azjAs known:
adopting a nonlinear optimization method to obtain the final product at the initial value
Figure 737884DEST_PATH_IMAGE015
For translation vector txj，tyj，tzjCalculating a partial derivative;
for is to
Figure 428497DEST_PATH_IMAGE016
Taylor expansion is carried out to obtain a vector t related to translationxj，tyj，tzjCorrection amount dt ofxj，dtyj，dtzjA system of linear equations of (c);
the linear equation set is solved by a least square method to obtain the correction dtxj，dtyj，dtzj；
According to the correction dtxj，dtyj，dtzjFor translation vector txj，tyj，tzjAnd (5) correcting:
Figure 993208DEST_PATH_IMAGE017
and obtaining a correction value.
Third, if only the rotation angle is corrected, axj，ayj，azjIs unknown, txj，tyj，tzjTo be known:
adopting a nonlinear optimization method to carry out the operation at an initial value
Figure 794680DEST_PATH_IMAGE018
For the rotation angle axj，ayj，azjCalculating a deviation derivative;
to pair
Figure 698919DEST_PATH_IMAGE019
Taylor expansion is performed to obtain the angle of rotation axj，ayj，azjCorrection amount daxj，dayj，dazjA system of linear equations of (c);
the linear equation set is solved by least square method to obtain correction daxj，dayj，dazj；
According to the correction amount daxj，dayj，dazjFor the rotation angle axj，ayj，azjAnd (3) correcting:
Figure 622751DEST_PATH_IMAGE020
and obtaining a correction value.
The first case and the third case occur more frequently, and the second case is less frequent because it is difficult to acquire attitude data (rotation angle) with high accuracy.
S405: after obtaining the correction value, the correction value is used as a new translation vector txj，tyj，tzjAnd/or angle of rotation axj，ayj，azjContinue to solve for newA correction value; and repeating the iteration initial value and solving a new correction value until the iteration initial value is repeated to the specified times and/or the error of the initial value calculation of the two times of updating is less than the specified threshold value. The initial value and the correction value can be continuously updated in the measuring process according to the requirement until the measurement is finished.
The designated times are 50-100. The threshold is specified to be 0.001-0.01.
By adopting the scheme, the relative position and the posture of each camera do not need to be accurately calibrated in advance, and the system is stable and the parameters are unchanged when the system is used; the method has the advantages that each camera is not required to be fixedly connected with a posture sensor, a control point target with accurately known coordinates is not required to be arranged in a measurement view field, only a simple scattered bright spot irradiator is required, the irradiation angle of each bright spot, the relation between the scattered bright spot irradiator and a stereoscopic vision camera and the like are not required to be known or calibrated, the stability of the scattered bright spot irradiator is not required to be kept in the using process, the interrelation of the irradiation angles of the bright spots is not required to be kept, the relative relation between the scattered bright spot irradiator and the stereoscopic vision camera is also not required to be kept, the relative position posture parameters between the cameras of the stereoscopic vision measurement system can be calibrated in real time, the precision of the stereoscopic vision measurement is ensured under the conditions of disturbance, change and the like of parameters of the stereoscopic vision measurement system, and meanwhile, the method can adapt to the conditions that the surface has no stable texture, the environmental illumination condition is poor, the motion of the measurement scene is changed, and the target points cannot be calibrated in the area of a target to be measured.
As a third embodiment of the present invention, there is provided a self-calibrating measurement system comprising the apparatus of the first embodiment and a stereo vision measurement system. The stereo vision measurement system calibrates the first camera C according to the relative position posture0And a second camera Cj。
As a fourth embodiment of the present invention, a method for measuring an object to be measured is provided, which employs the self-calibrated measuring system of the third embodiment.
The method comprises the following steps:
calibrating a stereoscopic vision camera of the vision measuring system by using the method of the second embodiment; the object 1 to be measured is measured using a stereo vision camera.
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes will occur to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (8)

1. A calibration method for the relative position and attitude of a stereo vision measurement system is characterized by comprising the following steps:
s1: the stereovision measurement system is used for shooing the target that awaits measuring, and the stereovision measurement system includes m stereovision cameras, and m is the natural number more than or equal to 2, and m stereovision cameras include: a first camera and m-1 second cameras; shooting a target to be detected from different angles by m stereoscopic vision cameras; projecting n bright spots to a target to be detected, wherein the n bright spots are all in the shooting range of a stereoscopic vision camera; n is a natural number greater than or equal to 3;
s2: all stereoscopic vision cameras shoot a target image to be detected with n bright spots, image points corresponding to the n bright spots in the target image to be detected are extracted, and image points of a second camera and image points of a first camera are matched with image points of the second camera in a same name mode;
s3: calculating the normalized image coordinates of the image points of the stereoscopic vision camera;
s4: respectively solving the relative position posture of each second camera and the first camera according to the image points with the same name and the normalized image coordinates of the image points of the first camera and the second camera;
in step S4, the method for solving the relative position and posture of each second camera and the first camera according to the image point with the same name and the normalized image coordinates of the image points of the first camera and the second camera includes:
setting an initial value of the relative position posture of the camera;
establishing a relative position attitude equation according to the relation between the first camera and the second camera;
listing a relative position attitude equation according to all the image points with the same name;
solving a relative position attitude equation according to the initial value and the listed relative position attitude equation to obtain a corrected value;
the relative position attitude equation is:
Figure 274685DEST_PATH_IMAGE002
；
the second camera and the first camera relative position pose comprises a translation vector and/or a rotation angle;
wherein the translation vector is
Figure 654106DEST_PATH_IMAGE003
The rotation matrix is
Figure 441015DEST_PATH_IMAGE004
At a rotation angle of
Figure 108714DEST_PATH_IMAGE005
The rotation matrix is a trigonometric combination of rotation angles; x, y, z are mutually perpendicular axes, txj、tyj、tzjRepresenting the length of movement of the camera in three axes, azj、ayj、axjThe angle of the camera rotating around the three axes of z, y and x in turn, r0j=cosayj×cosazj、r1j=sinaxj×sinayj×cosazj-cosaxj×sinazj、r2j=cosaxj×sinayj×cosazj+sinaxj×sinaxzj、r3j=cosayj×sinazj、r4j=sinaxj×sinayj×sinazj+cosaxj×cosazj、r5j=cosaxj×sinayj×sinazj-sinaxj×cosazj、r6j=-sinayj、r7j=sinaxj×cosayj、r8j=cosaxj×cosayj。
2. The method for calibrating the relative position and orientation of the stereo vision measurement system according to claim 1, wherein in step S3, the method for calculating the normalized image coordinates of the image point of the stereo vision camera comprises:
calculating the normalized image coordinates of the image points according to the internal parameters of the stereoscopic vision camera;
the internal parameters of the stereoscopic vision camera comprise principal point coordinates and equivalent focal length;
the image point normalized image coordinates of the first camera are:
Figure 514203DEST_PATH_IMAGE006
；
the image point normalized image coordinates of the second camera are:
Figure 147528DEST_PATH_IMAGE007
；
wherein the principal point coordinate of the first camera is (C x0 , C y0 ) The equivalent focal length of the first camera is (F x0 , F y0 ) The principal point coordinate of the second camera is (C xj , C yj ) The equivalent focal length of the second camera is (F xj , F yj )，j=1,2,...,m-1；i=0,1，...，n-1。
3. The method for calibrating the relative position and orientation of the stereo vision measurement system of claim 1, wherein the method for separately solving the relative position and orientation of each second camera with respect to the first camera according to the image coordinates of the same name image point and the image points of the first camera and the second camera further comprises: after the correction value is obtained, the correction value is used as a new initial value to continuously solve a new correction value; and repeating the iteration initial value and solving a new correction value until the iteration initial value is repeated until the specified times and/or the error of the initial value calculation of the two times of updating is less than the specified threshold value or the measurement of the target to be measured is finished.
4. The stereo vision measurement system relative position and orientation calibration method of claim 1, wherein the method for solving the relative position and orientation equation according to the initial value and the listed relative position and orientation equation comprises:
adopting a nonlinear optimization method to carry out the operation at an initial value
Figure DEST_PATH_IMAGE008
Calculating a deviation of the translation vector and/or the rotation angle;
to pair
Figure 781639DEST_PATH_IMAGE009
Performing Taylor expansion to obtain a linear equation set of correction quantity related to the translation vector and/or the rotation angle;
solving the linear equation set by a least square method to obtain a correction quantity;
and correcting the translation vector and/or the rotation angle according to the correction amount to obtain a correction value.
5. A method for measuring an object to be measured, characterized in that a stereoscopic vision camera of a stereoscopic vision measuring system is calibrated by using the calibration method for the relative position and posture of the stereoscopic vision measuring system according to any one of claims 1 to 4;
and measuring the object to be measured by using a stereoscopic vision camera.
6. A relative position posture calibration device of a stereo vision measurement system is characterized by comprising:
a scattered bright spot irradiator capable of irradiating n bright spots to a target to be measured, wherein n is a natural number greater than or equal to 3;
the homonymous image point matching module is used for extracting image points corresponding to the n bright points in the target image to be detected and respectively matching the image points of the second camera with the image points of the first camera;
the image coordinate calculation module is used for calculating the normalized image coordinate of the image point of the stereoscopic vision camera;
the relative position and posture calculation module is used for respectively solving the relative position and posture of each second camera and the first camera according to the image points with the same name and the normalized image coordinates of the image points of the first camera and the second camera;
wherein, the stereovision measurement system is used for shooing the target that awaits measuring, and the stereovision measurement system includes m stereovision cameras, and m is more than or equal to 2's natural number, and m stereovision cameras include: a first camera and m-1 second cameras; shooting a target to be detected from different angles by m stereoscopic vision cameras;
the method for respectively solving the relative position and the posture of each second camera and the first camera according to the image points with the same name and the normalized image coordinates of the image points of the first camera and the second camera comprises the following steps:
setting an initial value of the relative position posture of the camera;
establishing a relative position attitude equation according to the relation between the first camera and the second camera;
listing a relative position attitude equation according to all the image points with the same name;
solving a relative position attitude equation according to the initial value and the listed relative position attitude equation to obtain a corrected value;
the relative position attitude equation is as follows:
Figure 375782DEST_PATH_IMAGE011
；
the second camera and the first camera relative position pose comprises a translation vector and/or a rotation angle;
wherein the translation vector is
Figure 333955DEST_PATH_IMAGE003
The rotation matrix is
Figure 87366DEST_PATH_IMAGE004
At a rotation angle of
Figure 419340DEST_PATH_IMAGE005
The rotation matrix is a trigonometric combination of rotation angles; x, y, z are mutually perpendicular axes, txj、tyj、tzjRepresents the length of movement of the camera in three axial directions, azj、ayj、axjThe angle of the camera rotating around the three axes of z, y and x in turn, r0j=cosayj×cosazj、r1j=sinaxj×sinayj×cosazj-cosaxj×sinazj、r2j=cosaxj×sinayj×cosazj+sinaxj×sinaxzj、r3j=cosayj×sinazj、r4j=sinaxj×sinayj×sinazj+cosaxj×cosazj、r5j=cosaxj×sinayj×sinazj-sinaxj×cosazj、r6j=-sinayj、r7j=sinaxj×cosayj、r8j=cosaxj×cosayj。
7. The stereo vision measurement system relative position and orientation calibration device of claim 6, wherein the dispersed bright spot irradiator is a laser.
8. A self-calibrating measurement system, comprising:
the stereo vision measurement system relative position and orientation calibration device and the stereo vision measurement system according to claim 6 or 7.
CN202210891478.2A 2022-07-27 2022-07-27 Position and attitude calibration method and device, and method and system for measuring target to be measured Active CN114964316B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210891478.2A CN114964316B (en) 2022-07-27 2022-07-27 Position and attitude calibration method and device, and method and system for measuring target to be measured

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210891478.2A CN114964316B (en) 2022-07-27 2022-07-27 Position and attitude calibration method and device, and method and system for measuring target to be measured

Publications (2)

Publication Number Publication Date
CN114964316A CN114964316A (en) 2022-08-30
CN114964316B true CN114964316B (en) 2022-11-01

Family

ID=82968762

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210891478.2A Active CN114964316B (en) 2022-07-27 2022-07-27 Position and attitude calibration method and device, and method and system for measuring target to be measured

Country Status (1)

Country Link
CN (1) CN114964316B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116753918B (en) * 2023-06-19 2024-03-19 中国人民解放军61540部队 Ground target position estimation method and device based on empty antenna array sensor

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107255443A (en) * 2017-07-14 2017-10-17 北京航空航天大学 Binocular vision sensor field calibration method and device under a kind of complex environment
JP2019052983A (en) * 2017-09-15 2019-04-04 キヤノン株式会社 Calibration method and calibrator
CN110296691A (en) * 2019-06-28 2019-10-01 上海大学 Merge the binocular stereo vision measurement method and system of IMU calibration
CN110332887A (en) * 2019-06-27 2019-10-15 中国地质大学（武汉） A kind of monocular vision pose measurement system and method based on characteristic light punctuate
CN111220120A (en) * 2019-12-02 2020-06-02 上海航天控制技术研究所 Moving platform binocular ranging self-calibration method and device
CN111784778A (en) * 2020-06-04 2020-10-16 华中科技大学 Binocular camera external parameter calibration method and system based on linear solving and nonlinear optimization
CN113240740A (en) * 2021-05-06 2021-08-10 四川大学 Attitude measurement method based on phase-guided binocular vision dense marking point matching
CN113538598A (en) * 2021-07-21 2021-10-22 北京能创科技有限公司 Active stereo vision system calibration method
CN113739765A (en) * 2021-08-23 2021-12-03 中国人民解放军63660部队 Binocular collaborative drop point measurement method without additional control point
WO2022120567A1 (en) * 2020-12-08 2022-06-16 深圳先进技术研究院 Automatic calibration system based on visual guidance
CN114705122A (en) * 2022-04-13 2022-07-05 成都飞机工业（集团）有限责任公司 Large-field stereoscopic vision calibration method

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109074664A (en) * 2017-10-26 2018-12-21 深圳市大疆创新科技有限公司 Posture scaling method, equipment and unmanned vehicle
JP2022089269A (en) * 2020-12-04 2022-06-16 株式会社日立製作所 Calibration device and calibration method

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107255443A (en) * 2017-07-14 2017-10-17 北京航空航天大学 Binocular vision sensor field calibration method and device under a kind of complex environment
JP2019052983A (en) * 2017-09-15 2019-04-04 キヤノン株式会社 Calibration method and calibrator
CN110332887A (en) * 2019-06-27 2019-10-15 中国地质大学（武汉） A kind of monocular vision pose measurement system and method based on characteristic light punctuate
CN110296691A (en) * 2019-06-28 2019-10-01 上海大学 Merge the binocular stereo vision measurement method and system of IMU calibration
CN111220120A (en) * 2019-12-02 2020-06-02 上海航天控制技术研究所 Moving platform binocular ranging self-calibration method and device
CN111784778A (en) * 2020-06-04 2020-10-16 华中科技大学 Binocular camera external parameter calibration method and system based on linear solving and nonlinear optimization
WO2022120567A1 (en) * 2020-12-08 2022-06-16 深圳先进技术研究院 Automatic calibration system based on visual guidance
CN113240740A (en) * 2021-05-06 2021-08-10 四川大学 Attitude measurement method based on phase-guided binocular vision dense marking point matching
CN113538598A (en) * 2021-07-21 2021-10-22 北京能创科技有限公司 Active stereo vision system calibration method
CN113739765A (en) * 2021-08-23 2021-12-03 中国人民解放军63660部队 Binocular collaborative drop point measurement method without additional control point
CN114705122A (en) * 2022-04-13 2022-07-05 成都飞机工业（集团）有限责任公司 Large-field stereoscopic vision calibration method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Estimating Relative Pose between Nonoverlapping Cameras by Four Laser Pointers Based on General Camera Model;Shigang Li 等;《2017 4th IAPR Asian Conference on Pattern Recognition (ACPR)》;20181216;第1-5页 *
基于共面圆的距离传感器与相机的相对位姿标定;王硕 等;《自动化学报》;20200630;第46卷(第06期);第1154-1165页 *

Also Published As

Publication number Publication date
CN114964316A (en) 2022-08-30

Similar Documents

Publication Publication Date Title
CN110296691B (en) IMU calibration-fused binocular stereo vision measurement method and system
CN109029299B (en) Dual-camera measuring device and method for butt joint corner of cabin pin hole
Luna et al. Calibration of line-scan cameras
Luhmann Precision potential of photogrammetric 6DOF pose estimation with a single camera
Boochs et al. Increasing the accuracy of untaught robot positions by means of a multi-camera system
CN100376883C (en) Pixel frequency based star sensor high accuracy calibration method
US20220230348A1 (en) Method and apparatus for determining a three-dimensional position and pose of a fiducial marker
CN111915685B (en) Zoom camera calibration method
CN109974618B (en) Global calibration method of multi-sensor vision measurement system
CN110703230A (en) Position calibration method between laser radar and camera
CN114964316B (en) Position and attitude calibration method and device, and method and system for measuring target to be measured
CN112229323B (en) Six-degree-of-freedom measurement method of checkerboard cooperative target based on monocular vision of mobile phone and application of six-degree-of-freedom measurement method
Ding et al. A robust detection method of control points for calibration and measurement with defocused images
CN115014398B (en) Monocular stereoscopic vision measurement system position and attitude calibration method, device and system
Yuan et al. A precise calibration method for line scan cameras
CN106959704B (en) Control method and system of three-dimensional topography measuring instrument
Ye et al. A calibration trilogy of monocular-vision-based aircraft boresight system
CN111754584A (en) Remote large-field-of-view camera parameter calibration system and method
KR102167847B1 (en) System and Method for Calibration of Mobile Mapping System Using Laser Observation Equipment
CN113483879B (en) Small satellite flutter high-speed video measurement method
Oniga et al. Metric and Non-Metric Cameras Calibration for the Improvement of Real-Time Monitoring Process Results.
Yang et al. Improved calibration method of binocular vision measurement system for large hot forging
CN113405532A (en) Forward intersection measuring method and system based on structural parameters of vision system
CN111595289A (en) Three-dimensional angle measurement system and method based on image processing
Zhang et al. Flexible on-orbit calibration for monocular camera and laser rangefinder integrated pose measurement system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant