Cubic calibration block-based common-view-field-free multi-vision positioning system external parameter calibration method

Abstract

The invention discloses a common-view-field-free multi-vision positioning system external parameter calibration method based on a cubic calibration block, and belongs to the technical field of precise vision positioning and automatic control. The problems that a calibration route of a multi-vision positioning system calibration method without a common view field depends on an intermediary vision instrument, a large-size reference or a submillimeter-level repeated positioning requirement for a transfer mechanism, the hardware investment is high, the field layout is complex, the error chain is long, and the external parameter calibration accuracy is low are solved. According to the method, the physical constraint of calibration block synchronous pose change is converted into a linear equation set problem which can be stably solved through mathematical optimization, so that an external parameter matrix between two cameras without a common view field is accurately solved. The method is mainly applied to high-precision pose monitoring.

Description

Technical Field

[0001] This invention belongs to the field of precision vision positioning and automatic control technology. Background Technology

[0002] In high-end applications such as advanced manufacturing, aerospace, and large scientific facilities, multiple visual positioning systems are often deployed within a spatial range of tens or even hundreds of meters to achieve high-precision six-degree-of-freedom pose monitoring of large components or moving targets throughout their entire journey. The cameras in different visual positioning systems are spaced far apart and lack a common field of view. Existing calibration routes either rely on intermediate visual instruments, require the creation of large-size references matching the coverage area, or impose sub-millimeter-level repeatability requirements on the transfer mechanism. This results in high hardware investment, complex on-site deployment, and long error chains, leading to low accuracy in extrinsic parameter calibration. These problems urgently need to be addressed. Summary of the Invention

[0003] The purpose of this invention is to address the problem that calibration methods for multi-vision positioning systems without a common field of view rely on intermediate vision instruments, large-size references, or impose sub-millimeter-level repeatability requirements on the transfer mechanism, resulting in high hardware investment, complex on-site deployment, and long error chains, leading to low accuracy in extrinsic parameter calibration. This invention provides an extrinsic parameter calibration method for multi-vision positioning systems without a common field of view based on cubic calibration blocks.

[0004] A method for extrinsic parameter calibration of a multi-view positioning system without a common field of view based on a cube calibration block is implemented using two multi-view visual positioning systems A and B. The method includes:

[0005] S1, Calibration phase of reference camera P1 in multi-view visual positioning system A:

[0006] Place the cube calibration block within the field of view of the multi-view visual positioning system A. The pose of the cube calibration block is fine-tuned, and an image is taken once using the reference camera P1 of the multi-view visual positioning system A for each pose, clearly capturing one image of the corresponding calibration face of the cube calibration block; using the obtained image... From the images, we obtain the extrinsic parameter matrix of the coordinate system CamA of the reference camera P1 relative to its calibration surface coordinate system C for each pose;

[0007] The calibration faces of the cubic calibration block are rigidly connected;

[0008] S2, Cube calibration block transfer stage: The cube calibration block, which is located within the field of view of the multi-view visual positioning system A, is transferred to the field of view of the multi-view visual positioning system B using the target transfer device.

[0009] S3, Calibration phase of reference camera P2 in multi-view visual positioning system B:

[0010] Within the field of view of the multi-view visual positioning system B, The pose of the cube calibration block is fine-tuned, and the pose adjusted each time is the same as the pose adjusted within the field of view of the multi-view visual positioning system A.

[0011] At each pose, a reference camera P2 is used to take one clear image of the corresponding calibration face of the cube calibration block; using the obtained image... From the images, we obtain the coordinate system CamB of the reference camera P2 relative to its calibration plane coordinate system for each pose. The extrinsic parameter matrix;

[0012] S4. Calibration of extrinsic parameter matrix between two cameras in different systems:

[0013] Based on all the extrinsic parameter matrices corresponding to the two reference cameras P1 and P2, the extrinsic parameter matrix between the coordinate system CamA of reference camera P1 and the coordinate system CamB of reference camera P2 in the two multi-view visual positioning systems is obtained. .

[0014] Preferably, in step S4, the extrinsic parameter matrix between the coordinate system CamA of the reference camera P1 and the coordinate system CamB of the reference camera P2 in the two multi-view visual positioning systems is obtained. The implementation method is as follows:

[0015] S41. Suppose there is a point with coordinates C in coordinate system C. The point in the coordinate system The coordinates in are Then the following transformation relationship equation exists:

[0016] ;

[0017] in, For the first The coordinate system CamB of the reference camera P2 in this pose is relative to its calibration plane coordinate system. The extrinsic matrix, For the first The extrinsic parameter matrix of the coordinate system CamA of the reference camera P1 relative to its calibration surface coordinate system C under this pose. ;

[0018] S42. According to the transformation relationship equation, the following coefficient relationship equation exists;

[0019] ;

[0020] in, For the first The coordinate system CamB of the reference camera P2 in this pose is relative to its calibration plane coordinate system. The extrinsic matrix, For the first The extrinsic parameter matrix of the coordinate system CamA of the reference camera P1 relative to its calibration surface coordinate system C under this pose. ;

[0021] S43. Transform the coefficient relationship equation to obtain the transformed equation:

[0022] ;

[0023] S44. Using optimization algorithms such as least squares, the system of equations formed by all the transformed equations is optimized and solved to obtain... .

[0024] Preferably, during the calibration phase of the reference camera P1 in the multi-view visual positioning system A, the obtained data is used... The method for obtaining the extrinsic parameter matrix of the coordinate system CamA of the reference camera P1 relative to its calibration surface coordinate system C under each pose is achieved by Zhang Zhengyou calibration method or PnP pose estimation method.

[0025] Preferably, during the calibration phase of the reference camera P2 in the multi-view visual positioning system B, the obtained data is used... From the images, we obtain the coordinate system CamB of the reference camera P2 relative to its calibration plane coordinate system for each pose. The extrinsic parameter matrix can be implemented using the Zhang Zhengyou calibration method or the PnP pose estimation method.

[0026] Preferably, the depths of field of the multi-view visual positioning systems A and B are the same or different.

[0027] Preferably, in step S1, The calibration plane captured in each pose is the same calibration plane.

[0028] Preferably, in step S3, The calibration plane captured in each pose is the same calibration plane.

[0029] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the extrinsic parameter calibration method for a multi-vision positioning system without a common field of view based on cube calibration blocks.

[0030] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the extrinsic parameter calibration method for a multi-vision positioning system without a common field of view based on cube calibration blocks.

[0031] The beneficial effects of this invention are:

[0032] The core effect of the extrinsic parameter calibration method for a multi-vision positioning system without a common field of view based on a cubic calibration block described in this invention is that it successfully transforms the physical constraint of "synchronous pose change of the calibration block" into a linear equation problem that can be stably solved through mathematical optimization, thereby accurately solving the extrinsic parameter matrix between two cameras without a common field of view.

[0033] This invention sequentially feeds the target (cube calibration block) into the fields of view of two systems and performs identical pose adjustment sequences at both ends. It eliminates the need to record movement trajectories, additional cameras, or lasers, and directly calculates inter-system extrinsic parameters using hand-eye closed-chain constraints. This method reduces the complexity of calibration hardware and sensitivity to transfer accuracy, offers high extrinsic parameter calibration accuracy, and is relatively simple to operate.

[0034] This invention utilizes a visual positioning system A to observe and locate a target (cube calibration block) and obtain its precise pose (i.e., extrinsic parameter matrix). Subsequently, a target delivery device delivers the target (cube calibration block) to the visual positioning system B according to a predetermined pose. Through this series of operations, it can be ensured that the initial position and attitude deviation of the target is controlled within a very small range when it enters the visual positioning system B, effectively avoiding the imaging blurring problem caused by excessive initial deviation, and improving the efficiency and reliability of the system's acquisition and alignment. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the positional relationship between two multi-view visual positioning systems A and B with different depths of field as described in this invention;

[0036] Figure 2 This is a three-dimensional structural diagram of the cube calibration block;

[0037] Figure 3 This is a schematic diagram of an orthogonal visual positioning system. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0040] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but this is not intended to limit the scope of the invention.

[0041] Specific Implementation Method 1: Combination Figure 1 and Figure 2 This embodiment describes the extrinsic parameter calibration method for a multi-view positioning system without a common field of view based on a cube calibration block. This method is implemented using two multi-view visual positioning systems, A and B, and includes:

[0042] S1, Calibration phase of reference camera P1 in multi-view visual positioning system A:

[0043] Place the cube calibration block within the field of view of the multi-view visual positioning system A. The pose of the cube calibration block is fine-tuned, and a picture is taken once by the reference camera P1 of the multi-view visual positioning system A in each pose, so as to clearly capture an image of the corresponding calibration face of the cube calibration block.

[0044] The calibration plane captured under all poses is the same calibration plane;

[0045] Utilize the obtained From the images, we obtain the extrinsic parameter matrix of the coordinate system CamA of the reference camera P1 relative to its calibration surface coordinate system C for each pose;

[0046] The calibration faces of the cubic calibration block are rigidly connected;

[0047] S2, Cube calibration block transfer stage: The cube calibration block, which is located within the field of view of the multi-view visual positioning system A, is transferred to the field of view of the multi-view visual positioning system B using a target delivery device (such as a robotic arm or translation stage).

[0048] S3, Calibration phase of reference camera P2 in multi-view visual positioning system B:

[0049] Within the field of view of the multi-view visual positioning system B, The pose of the cube calibration block is fine-tuned, and the pose adjusted each time is the same as the pose adjusted within the field of view of the multi-view visual positioning system A.

[0050] At each pose, a reference camera P2 is used to take a picture, clearly capturing an image of the corresponding calibration face of the cube calibration block;

[0051] The calibration plane captured under all poses is the same calibration plane;

[0052] Utilize the obtained From the images, we obtain the coordinate system CamB of the reference camera P2 relative to its calibration plane coordinate system for each pose. The extrinsic parameter matrix;

[0053] S4. Calibration of extrinsic parameter matrix between two cameras in different systems:

[0054] Based on all the extrinsic parameter matrices corresponding to the two reference cameras P1 and P2, the extrinsic parameter matrix between the coordinate system CamA of reference camera P1 and the coordinate system CamB of reference camera P2 in the two multi-view visual positioning systems is obtained. .

[0055] Clear shooting in application refers to the ability to identify the characteristic corner points or patterns of the calibration surface. Figure 1 middle, For calibration surface coordinate system The transformation matrix relative to the calibration plane coordinate system C is used in systems A and B using a cubic calibration block. Under the condition that the next pose adjustment is completely corresponding. It remains unchanged.

[0056] This implementation utilizes a cube calibration block as an "unchanging geometric reference" and, through a carefully designed process, establishes a precise coordinate transformation relationship between two spatially separated, non-overlapping multi-vision systems. First, within the field of view of system A, by repeatedly changing the cube's orientation, the transformation relationship between the reference camera coordinate system CamA of system A and the calibration block coordinate system C is precisely determined. Then, using a target delivery device, the entire cube calibration block is moved to the field of view of system B, and the previously synchronized multiple orientation changes are rigorously reproduced, thereby calibrating the transformation relationship between the reference camera coordinate system CamB of system B and the current calibration block coordinate system. The transformation relationship.

[0057] Two strong constraints:

[0058] First, there are the calibration block pose synchronization constraints, meaning that the attitude adjustment sequence within the field of view of system B is completely synchronized with the sequence within the field of view of system A. It is through these constraints that the calibration block coordinate system C and... Although they are located in different positions in space, their relative pose relationships are precisely maintained and transmitted;

[0059] Second, there is a rigid connection between the calibration surfaces of the calibration block. That is, reference camera P1 in system A and reference camera P2 in system B photograph the same or different calibration surfaces of the calibration block. Under the condition of synchronous pose constraints of the calibration block, calibration surface C and... The relative pose relationship between them remains unchanged.

[0060] Finally, by mathematically synthesizing the two transformation chains with the fixed pose relationships of the cameras in the two visual positioning systems, the extrinsic parameter matrix between the two camera coordinate systems CamA and CamB, which originally had no common field of view, can be solved, i.e., the rotation and translation relationship. This method cleverly avoids the traditional requirement of a large field of view or complex markers for a shared field of view. It achieves efficient and high-precision transfer of spatial reference by using a target delivery device and a standard geometric object (cube). It is particularly suitable for the integration and calibration of multi-station, multi-depth-of-field visual positioning systems with separate workspaces.

[0061] In practical applications, during the calibration phase of the reference camera P1 in the multi-view visual positioning system A, the obtained data is used... The extrinsic parameter matrix of the coordinate system CamA of the reference camera P1 relative to its calibration surface coordinate system C under each pose is obtained from the Zhang Zhengyou calibration method or the PnP pose estimation method. In the calibration stage of the reference camera P2 of the multi-view visual positioning system B, the obtained... From the images, we obtain the coordinate system CamB of the reference camera P2 relative to its calibration plane coordinate system for each pose. The extrinsic parameter matrix is ​​implemented using either the Zhang Zhengyou calibration method or the PnP pose estimation method. Furthermore, the extrinsic parameter matrix is ​​in the form of... ;

[0062] in, This is the rotation matrix between two coordinate systems. The translation vector between the two coordinate systems is given. Inverting this extrinsic parameter matrix yields the coordinates from the camera coordinate system CamB to the calibration surface coordinate system. The extrinsic parameter matrix is ​​in the form of: .

[0063] Both multi-view visual positioning systems A and B can be orthogonal visual positioning systems. Figure 3 A schematic diagram of an orthogonal vision positioning system is presented. Figure 3 In the figures, reference numeral 1 represents the upper CCD, reference numeral 2 represents the upper microscope objective, reference numeral 3 represents the middle CCD1, reference numeral 4 represents the middle microscope objective 1, reference numeral 5 represents the middle microscope objective 2, reference numeral 6 represents the middle CCD2, reference numeral 7 represents the lower CCD, and reference numeral 8 represents the lower microscope objective.

[0064] In practical applications, the depths of field of multi-view visual positioning systems A and B may be the same or different.

[0065] Further, see Figure 1 In step S4, the extrinsic parameter matrix between the coordinate system CamA of the reference camera P1 and the coordinate system CamB of the reference camera P2 in two multi-view visual positioning systems with different depths of field is obtained. The implementation method is as follows:

[0066] S41. Suppose there is a point with coordinates C in coordinate system C. The point in the coordinate system The coordinates in are Then the following transformation relationship equation exists:

[0067] ;

[0068] in, For the first The coordinate system CamB of the reference camera P2 in this pose is relative to its calibration plane coordinate system. The extrinsic matrix, For the first The extrinsic parameter matrix of the coordinate system CamA of the reference camera P1 relative to its calibration surface coordinate system C under this pose. ;

[0069] S42. According to the transformation relationship equation, the following coefficient relationship equation exists;

[0070] ;

[0071] ;

[0072]

[0073] ;

[0074] That is:

[0075] ;

[0076] in, For the first The coordinate system CamB of the reference camera P2 in this pose is relative to its calibration plane coordinate system. The extrinsic matrix, For the first The extrinsic parameter matrix of the coordinate system CamA of the reference camera P1 relative to its calibration surface coordinate system C under this pose. ;

[0077] S43. Transform the coefficient relationship equation to obtain the transformed equation:

[0078] ;

[0079] S44, due to It can be equivalent to a typical The problem is that, therefore, optimization algorithms such as least squares are used to solve all the transformed equations ( The system of equations (formed by the equations) is optimized and solved to obtain the following results. .

[0080] In this preferred embodiment, by constructing a complete coordinate transformation chain from CamA to CamB and introducing equivalent expressions for the same point on the calibration block in different camera coordinate systems, a homogeneous equation with the desired extrinsic parameter matrix as the core unknown is derived. By repeating this process under different synchronous poses, a set of equations can be accumulated. The construction method of this set of equations cleverly reduces complex spatial transformation relationships to the direct solution of a single transformation matrix. Finally, by using optimization algorithms such as least squares to solve this set of equations, random errors in a single measurement can be effectively suppressed, thereby robustly and accurately calculating the rotation and translation relationship between the two camera coordinate systems, i.e., the extrinsic parameter matrix.

[0081] Specific Embodiment Two: An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the extrinsic parameter calibration method for a multi-vision positioning system without a common field of view based on a cube calibration block as described in Specific Embodiment One.

[0082] Specific Implementation Method 3: A computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the extrinsic parameter calibration method for a multi-vision positioning system without a common field of view based on a cube calibration block as described in Specific Implementation Method 1.

[0083] While the invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the invention. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that different dependent claims and features described herein can be combined in ways different from those described in the original claims. It is also understood that features described in conjunction with individual embodiments can be used in other described embodiments.

Claims

1. A method for extrinsic parameter calibration of a multi-vision positioning system without a common field of view based on cubic calibration blocks, characterized in that, Based on two multi-view visual positioning systems A and B, this method includes: S1, Calibration phase of reference camera P1 in multi-view visual positioning system A: Place the cube calibration block within the field of view of the multi-view visual positioning system A. The pose of the cube calibration block is fine-tuned, and an image is taken once using the reference camera P1 of the multi-view visual positioning system A for each pose, clearly capturing one image of the corresponding calibration face of the cube calibration block; using the obtained image... From the images, we obtain the extrinsic parameter matrix of the coordinate system CamA of the reference camera P1 relative to its calibration surface coordinate system C for each pose; The calibration faces of the cubic calibration block are rigidly connected; S2, Cube calibration block transfer stage: The cube calibration block, which is located within the field of view of the multi-view visual positioning system A, is transferred to the field of view of the multi-view visual positioning system B using the target transfer device. S3, Calibration phase of reference camera P2 in multi-view visual positioning system B: Within the field of view of the multi-view visual positioning system B, The pose of the cube calibration block is fine-tuned, and the pose adjusted each time is the same as the pose adjusted within the field of view of the multi-view visual positioning system A. At each pose, a reference camera P2 is used to take one clear image of the corresponding calibration face of the cube calibration block; using the obtained image... From the images, we obtain the coordinate system CamB of the reference camera P2 relative to its calibration plane coordinate system for each pose. The extrinsic parameter matrix; S4. Calibration of extrinsic parameter matrix between two cameras in different systems: Based on all the extrinsic parameter matrices corresponding to the two reference cameras P1 and P2, the extrinsic parameter matrix between the coordinate system CamA of reference camera P1 and the coordinate system CamB of reference camera P2 in the two multi-view visual positioning systems is obtained. .

2. The extrinsic parameter calibration method for a multi-vision positioning system without a common field of view based on a cube calibration block according to claim 1, characterized in that, In step S4, the extrinsic parameter matrix between the coordinate system CamA of reference camera P1 and the coordinate system CamB of reference camera P2 in the two multi-view visual positioning systems is obtained. The implementation method is as follows: S41. Suppose there is a point with coordinates C in coordinate system C. The point is in the coordinate system The coordinates in are Then the following transformation relationship equation exists: ； in, For the first The coordinate system CamB of the reference camera P2 in this pose is relative to its calibration plane coordinate system. The extrinsic matrix, For the first The extrinsic parameter matrix of the coordinate system CamA of the reference camera P1 relative to its calibration surface coordinate system C under this pose. ; S42. According to the transformation relationship equation, the following coefficient relationship equation exists; ； in, For the first The coordinate system CamB of the reference camera P2 in this pose is relative to its calibration plane coordinate system. The extrinsic matrix, For the first The extrinsic parameter matrix of the coordinate system CamA of the reference camera P1 relative to its calibration surface coordinate system C under this pose. ; S43. Transform the coefficient relationship equation to obtain the transformed equation: ； S44. Using optimization algorithms such as least squares, the system of equations formed by all the transformed equations is optimized and solved to obtain... .

3. The extrinsic parameter calibration method for a multi-vision positioning system without a common field of view based on a cube calibration block according to claim 1, characterized in that, During the calibration phase of the reference camera P1 in the multi-view visual positioning system A, the obtained data is used... The method for obtaining the extrinsic parameter matrix of the coordinate system CamA of the reference camera P1 relative to its calibration surface coordinate system C under each pose is achieved by Zhang Zhengyou calibration method or PnP pose estimation method.

4. The extrinsic parameter calibration method for a multi-vision positioning system without a common field of view based on a cube calibration block according to claim 1, characterized in that, During the calibration phase of the reference camera P2 in the multi-view visual positioning system B, the obtained data is used... From the images, we obtain the coordinate system CamB of the reference camera P2 relative to its calibration plane coordinate system for each pose. The extrinsic parameter matrix can be implemented using the Zhang Zhengyou calibration method or the PnP pose estimation method.

5. The extrinsic parameter calibration method for a multi-vision positioning system without a common field of view based on a cube calibration block according to claim 1, characterized in that, Multi-view visual positioning systems A and B may have the same or different depths of field.

6. The extrinsic parameter calibration method for a multi-vision positioning system without a common field of view based on a cube calibration block according to claim 1, characterized in that, In step S1, The calibration plane captured in each pose is the same calibration plane.

7. The extrinsic parameter calibration method for a multi-vision positioning system without a common field of view based on a cube calibration block according to claim 1, characterized in that, In step S3, The calibration plane captured in each pose is the same calibration plane.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the extrinsic parameter calibration method for a multi-vision positioning system without a common field of view based on a cube calibration block as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the extrinsic parameter calibration method for a multi-vision positioning system without a common field of view based on a cube calibration block as described in any one of claims 1 to 7.

Images