CN116952132A - Partition calibration method and system for multi-vision measurement - Google Patents
Partition calibration method and system for multi-vision measurement Download PDFInfo
- Publication number
- CN116952132A CN116952132A CN202310929701.2A CN202310929701A CN116952132A CN 116952132 A CN116952132 A CN 116952132A CN 202310929701 A CN202310929701 A CN 202310929701A CN 116952132 A CN116952132 A CN 116952132A
- Authority
- CN
- China
- Prior art keywords
- cameras
- camera
- cluster
- optimization
- parameters
- 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.)
- Pending
Links
- 238000005259 measurement Methods 0.000 title claims abstract description 43
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000005192 partition Methods 0.000 title claims abstract description 13
- 238000005457 optimization Methods 0.000 claims description 24
- 238000004590 computer program Methods 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 4
- 238000000638 solvent extraction Methods 0.000 claims description 4
- 238000003384 imaging method Methods 0.000 claims description 3
- 238000009434 installation Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000013316 zoning Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
Classifications
-
- 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
- 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
- G01B11/005—Measuring arrangements characterised by the use of optical techniques for measuring two or more coordinates coordinate measuring machines
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
- G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Multimedia (AREA)
- Probability & Statistics with Applications (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The application discloses a partition calibration method and a partition calibration system for multi-vision measurement, which are used for classifying a plurality of cameras to obtain a plurality of camera clusters; for each camera class cluster a, perform: the application optimizes the camera parameters in the camera cluster A to ensure that the whole camera parameters in the camera cluster A are optimal.
Description
Technical Field
The application relates to the field of multi-vision measurement, in particular to a zonal calibration method and a zonal calibration system for multi-vision measurement.
Background
In industrial measurement, most of the requirements based on high precision are that a three-coordinate measuring instrument is used for sampling and measuring a workpiece to be measured, and because three coordinates need to be moved to measure each point to be measured, the time cost of the workpiece with more points to be measured in a larger size is huge, and quick measurement is difficult to achieve.
With the development of the three-dimensional vision measurement and multi-vision measurement fields, a plurality of cameras distributed around a workpiece to be measured are utilized to take pictures at the same time, and three-dimensional reconstruction is carried out through calibrated cameras, so that the rapid measurement of the workpiece can be carried out. The quality of the calibration of the camera parameters is a decisive factor affecting the measurement accuracy, wherein the conventional multi-vision calibration method based on the hole center of the workpiece, as described in CN202110912869.3, is to obtain the image data of the calibration plates under the view angles of a plurality of cameras for initial calibration to obtain the corresponding internal parameters and external parameters of the cameras and distortion parameters, but the calibrated camera parameters are difficult to meet the measurement accuracy requirement without optimizing additional calibration parameters.
The secondary optimization based on camera calibration parameters can utilize the corresponding CAD digital analog of the workpiece and the corresponding numerical value measured by three coordinates as optimization targets to adjust the calibrated internal and external parameters of the camera, so as to achieve the requirement of high-precision measurement, but the workpiece to be measured in actual measurement easily does not achieve the requirement of local measurement precision, and if all the internal and external parameters of the camera are used as optimization parameters to be optimized, the global optimization is easily achieved, but the precision requirement of the local areas cannot be achieved.
Disclosure of Invention
The application provides a partition calibration method and a partition calibration system for multi-vision measurement, which are used for solving the technical problem that the calibration method for multi-vision measurement cannot consider global measurement precision and local measurement precision.
In order to solve the technical problems, the technical scheme provided by the application is as follows:
a partitioning calibration method for multi-vision measurement comprises the following steps:
classifying the cameras to obtain a plurality of camera clusters;
for each camera class cluster a, perform:
and optimizing the camera parameters in the camera cluster A to ensure that the whole camera parameters in the camera cluster A are optimal.
Preferably, the plurality of cameras are classified, including any one or a combination of several of the following classification methods:
the method comprises the following steps: dividing the shooting objects according to the distribution areas, and dividing the cameras of which the shooting objects are divided into the same partition into the same class;
the second method is as follows: classifying cameras according to the installation positions of the cameras;
and a third method: the distance between the cameras is used as a clustering variable to cluster the cameras, the cameras in the same cluster are of the same category, and for any two cameras A, B, the distance between the cameras A, B is the average re-projection error of the cameras A, B.
Preferably, the shooting object is a measuring hole of the workpiece to be measured.
Preferably, the average re-projection error of the camera A, B is calculated by:
and (3) performing BA optimization on the observation points shot by the cameras A, B together, solving a BA optimized solution by using an LM algorithm, and solving to obtain an optimized average re-projection error.
Preferably, the BA optimization formula is:
wherein C represents the camera's collection, P represents all 3D points, K i And [ R ] i |t i ]Respectively represent the internal parameters and external parameters of the ith camera in the first collection C, and 3D point X j At camera C i The re-projection error of (2) is that the 3D point passes through the corresponding projection formula K i ·[R i |t i ]·X j And actually its observation point x on the camera imaging ij Euclidean distance difference between them.
Preferably, the cameras are clustered, specifically:
and constructing a distance matrix among all cameras, and clustering all cameras by adopting a clustering algorithm.
Preferably, the camera parameters in the same camera cluster are optimized uniformly, specifically:
and performing global optimization on the camera parameters in the same camera cluster by using a BA optimization algorithm, and solving a solution of the BA optimization of the camera parameters in the same camera cluster by using an LM algorithm.
A computer system comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the method described above when the computer program is executed by the processor.
The application has the following beneficial effects:
1. according to the partition calibration method and the partition calibration system for the multi-vision measurement, the cameras are partitioned, global optimization is performed on each partitioned camera cluster, and the local measurement precision of the multi-vision measurement is considered while the global measurement precision of the multi-vision measurement is ensured.
2. In the preferred scheme, the application not only provides a relatively simple and manually feasible scheme such as a camera layout area, a hole measurement and the like, but also provides a method for clustering and automatically classifying and calibrating cameras by taking the finally optimized average re-projection error between two cameras as a distance.
In addition to the objects, features and advantages described above, the present application has other objects, features and advantages. The application will be described in further detail with reference to the accompanying drawings.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the application. In the drawings:
fig. 1 is a schematic view of camera partition provided in an embodiment of the present application;
FIG. 2 is a diagram showing an average re-projection error of any two phases provided by an embodiment of the present application;
FIG. 3 is a distance matrix diagram provided by an embodiment of the present application;
FIG. 4 is a schematic clustering diagram provided by an embodiment of the present application;
FIG. 5 is a flow chart of a zonal calibration method for multi-vision measurement provided by an embodiment of the present application.
Detailed Description
Embodiments of the application are described in detail below with reference to the attached drawings, but the application can be implemented in a number of different ways, which are defined and covered by the claims.
Embodiment one:
currently, in a workpiece hole center measurement scene, as shown in fig. 1, a plurality of 2D cameras are installed in a measurement chamber, then internal and external parameters of the cameras are optimized by using BA optimization, and the BA optimization is commonly solved by Levenberg-Marquardt (LM).
Different camera layouts can be provided for different workpieces, so that the cameras can cover all areas to be measured, wherein specific setting methods and calibration methods of the cameras can be seen in the schemes provided by CN202210017796.6, namely a calibration method and device for multi-vision measurement, a storage medium and camera equipment, and CN202110912869.3, namely a method, a device, a medium and a system for multi-vision calibration based on the hole center of the workpiece.
Because the number of cameras is large, the optimization is to minimize the re-projection error of the 3D observation points of all cameras, and the internal and external parameters of the local cameras are not optimized to be the best.
In order to solve the above-mentioned problems, as shown in fig. 5, the present embodiment provides a zonal calibration method for multi-vision measurement, which classifies cameras according to a desired camera distribution area or by using measurement holes.
As shown in fig. 1, the cameras in the left frame are one type, the cameras in the right frame are one type, and then BA optimization is performed on the two types of cameras to obtain corresponding camera parameters, where it is certain that the measurement points are photographed by the cameras in the left frame and the cameras in the right frame, and at this time, the camera parameters of the overlapping area can be optimized by using two types of cameras to reconstruct the hole and then averaging.
Of course, the cameras capable of shooting all holes in the left frame can be classified into one type by classifying according to the measuring holes, taking fig. 1 as an example, and the cameras capable of shooting holes in the right frame can be classified into another type, so that one camera can possibly use the two sets of external parameters according to the measuring points for maintenance of the camera, namely, the first type and the second type.
Of course, the above two kinds of partition calibration can be realized, but subjective factors are relatively large, and the optimal solution can be achieved without data support and classification. The patent provides a clustering method according to the distance between the cameras, and corresponding camera classification is automatically generated according to the input required category (hereinafter referred to as dynamic classification method)
The dynamic classification method comprises the following specific implementation steps:
s1, optimizing all cameras through the following BA optimization formula, and calculating the distance between any two cameras based on the optimization result.
Wherein, the formula corresponding to BA optimization is as follows:
wherein C represents the camera collection, P represents all 3D points, K i And [ R ] i |t i ]Respectively represent the internal parameters and external parameters of the ith camera in the first collection C, and 3D point X j (actual coordinates) at camera C i The re-projection error of (2) is that the 3D point passes through the corresponding projection formula K i ·[R i |t i ]·X j And actually its observation point x on the camera imaging ij Euclidean distance difference between (image coordinates).
By processing the set C, only two cameras and the observation points commonly photographed by the two cameras are randomly selected for BA optimization by using the LM algorithm, and the final average re-projection error is kept to be optimized, as 8.786937e-01 shown in fig. 2 is the final error, and this value is regarded as the "distance" between the two cameras.
S2, obtaining the 'distances' between every two cameras, and constructing a corresponding distance matrix as shown in fig. 3, wherein the distance between the two cameras is expressed by 1000 and then the 'distance' between the cameras and the camera is expressed by 0 because the two cameras do not necessarily have a common observation point (measuring point).
S3, inputting the category to be clustered by using a clustering method such as K-media and the like, and automatically classifying each camera. As shown in fig. 4, the corresponding classification result can be automatically obtained according to the corresponding clustering method and different clustering numbers, the abscissa is the ID of the camera, the ordinate is the category, the left graph is the result obtained by automatic clustering when the input category number is 2, and the right graph is the result obtained when the category number is 4.
The specific steps of the K-media clustering method are as follows:
step1, determining the value of the number K of clusters;
step2, inputting clustered data;
step3, randomly selecting K points in the data to serve as centroids;
step4, calculating the distance between other points and the mass center, and classifying the mass center closest to the mass center into a group to obtain clusters;
step5, calculating a distance once by taking each point between clusters as a centroid, and then selecting a point with the smallest distance as a new centroid;
step6, repeating Step4 and Step5 until the centroid is no longer offset;
the above steps can be expressed by the following formula, wherein C i Is the mass center, P i Is a non-centroid, so c is the sum of each non-centroid to centroid distance, so K-media is to minimize c, i.e., min (c)
The clustering method is various, the patent is not limited to a certain clustering method, other clustering algorithms such as K-MEANS algorithm K-MEDOIDS algorithm, CLARANS algorithm and other dividing algorithms, and density algorithms such as DBSCAN algorithm, OPTICS algorithm and DENCLUE algorithm are also applicable.
In addition, in the present embodiment, there is also provided a computer system including a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the above method when executing the computer program.
In summary, the zoning calibration method and system for multi-vision measurement in the application can ensure the global measurement precision of the multi-vision measurement and simultaneously consider the local measurement precision by zoning the cameras and globally optimizing the camera clusters in each zone.
The above description is only of the preferred embodiments of the present application and is not intended to limit the present application, but various modifications and variations can be made to the present application by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.
Claims (8)
1. The partition calibration method for the multi-vision measurement is characterized by comprising the following steps of:
classifying the cameras to obtain a plurality of camera clusters;
for each camera class cluster a, perform:
and optimizing the camera parameters in the camera cluster A to ensure that the whole camera parameters in the camera cluster A are optimal.
2. The method of claim 1, wherein classifying the plurality of cameras comprises any one or a combination of the following classification methods:
the method comprises the following steps: dividing the shooting objects according to the distribution areas, and dividing the cameras of which the shooting objects are divided into the same partition into the same class;
the second method is as follows: classifying cameras according to the installation positions of the cameras;
and a third method: the distance between the cameras is used as a clustering variable to cluster the cameras, the cameras in the same cluster are of the same category, and for any two cameras A, B, the distance between the cameras A, B is the average re-projection error of the cameras A, B.
3. The method for demarcating a region of a multiview measurement according to claim 2, wherein the photographic subject is a measuring hole of a workpiece to be measured.
4. The method of demarcating a region of a multiview measurement according to claim 2, wherein the average re-projection error of the camera A, B is calculated by:
and (3) performing BA optimization on the observation points shot by the cameras A, B together, solving a BA optimized solution by using an LM algorithm, and solving to obtain an optimized average re-projection error.
5. The method for partitioning calibration for multi-vision measurement of claim 4, wherein said BA optimization formula is:
wherein C represents the camera's collection, P represents all 3D points, K i And [ R ] i |t i ]Respectively represent the internal parameters and external parameters of the ith camera in the first collection C, and 3D point X j At camera C i The re-projection error of (2) is that the 3D point passes through the corresponding projection formula K i ·[R i |t i ]·X j And actually its observation point x on the camera imaging ij Euclidean distance difference between them.
6. The method for partitioning calibration for multi-vision measurement of claim 5, wherein the clustering of the cameras is specifically:
and constructing a distance matrix among all cameras, and clustering all cameras by adopting a clustering algorithm.
7. The method for partitioning calibration for multi-vision measurement according to any one of claims 1-6, wherein the unified optimization of camera parameters in the same camera cluster is specifically:
and performing global optimization on the camera parameters in the same camera cluster by using a BA optimization algorithm, and solving a solution of the BA optimization of the camera parameters in the same camera cluster by using an LM algorithm.
8. A computer system comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the method of any of the preceding claims 1 to 7 when the computer program is executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310929701.2A CN116952132A (en) | 2023-07-27 | 2023-07-27 | Partition calibration method and system for multi-vision measurement |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310929701.2A CN116952132A (en) | 2023-07-27 | 2023-07-27 | Partition calibration method and system for multi-vision measurement |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116952132A true CN116952132A (en) | 2023-10-27 |
Family
ID=88456246
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310929701.2A Pending CN116952132A (en) | 2023-07-27 | 2023-07-27 | Partition calibration method and system for multi-vision measurement |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116952132A (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103198524A (en) * | 2013-04-27 | 2013-07-10 | 清华大学 | Three-dimensional reconstruction method for large-scale outdoor scene |
| CN107966112A (en) * | 2017-12-03 | 2018-04-27 | 中国直升机设计研究所 | A kind of large scale rotor movement parameter measurement method |
| CN113516719A (en) * | 2021-04-21 | 2021-10-19 | 深圳臻像科技有限公司 | Camera calibration method, system and storage medium based on multiple homography matrixes |
| US20210329221A1 (en) * | 2020-04-21 | 2021-10-21 | Plato Systems, Inc. | Method and apparatus for camera calibration |
| CN114004901A (en) * | 2022-01-04 | 2022-02-01 | 南昌虚拟现实研究院股份有限公司 | Multi-camera calibration method and device, terminal equipment and readable storage medium |
| CN114399554A (en) * | 2021-12-08 | 2022-04-26 | 凌云光技术股份有限公司 | Calibration method and system of multi-camera system |
| CN114782542A (en) * | 2022-03-07 | 2022-07-22 | 合肥工业大学 | Annular multi-camera calibration method and system for panoramic measurement of large target |
| CN115564845A (en) * | 2022-10-12 | 2023-01-03 | 广东亿嘉和科技有限公司 | Regional binocular camera calibration method |
| CN116051652A (en) * | 2023-01-06 | 2023-05-02 | 浙江零跑科技股份有限公司 | Parameter calibration method, electronic equipment and storage medium |
-
2023
- 2023-07-27 CN CN202310929701.2A patent/CN116952132A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103198524A (en) * | 2013-04-27 | 2013-07-10 | 清华大学 | Three-dimensional reconstruction method for large-scale outdoor scene |
| CN107966112A (en) * | 2017-12-03 | 2018-04-27 | 中国直升机设计研究所 | A kind of large scale rotor movement parameter measurement method |
| US20210329221A1 (en) * | 2020-04-21 | 2021-10-21 | Plato Systems, Inc. | Method and apparatus for camera calibration |
| CN113516719A (en) * | 2021-04-21 | 2021-10-19 | 深圳臻像科技有限公司 | Camera calibration method, system and storage medium based on multiple homography matrixes |
| CN114399554A (en) * | 2021-12-08 | 2022-04-26 | 凌云光技术股份有限公司 | Calibration method and system of multi-camera system |
| CN114004901A (en) * | 2022-01-04 | 2022-02-01 | 南昌虚拟现实研究院股份有限公司 | Multi-camera calibration method and device, terminal equipment and readable storage medium |
| CN114782542A (en) * | 2022-03-07 | 2022-07-22 | 合肥工业大学 | Annular multi-camera calibration method and system for panoramic measurement of large target |
| CN115564845A (en) * | 2022-10-12 | 2023-01-03 | 广东亿嘉和科技有限公司 | Regional binocular camera calibration method |
| CN116051652A (en) * | 2023-01-06 | 2023-05-02 | 浙江零跑科技股份有限公司 | Parameter calibration method, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110738143B (en) | Positioning method and device, equipment and storage medium | |
| Park et al. | A look-up table based approach for solving the camera selection problem in large camera networks | |
| CN112634374B (en) | Stereoscopic calibration method, device and system for binocular camera and binocular camera | |
| Shortis et al. | Principal point behaviour and calibration parameter models for Kodak DCS cameras | |
| CN113592957B (en) | Multi-laser radar and multi-camera combined calibration method and system | |
| CN108648241A (en) | A kind of Pan/Tilt/Zoom camera field calibration and fixed-focus method | |
| CN112634373B (en) | Visual measurement system dynamic correction method based on zero-expansion ceramic calibration plate | |
| CN112085848A (en) | Method and equipment for optimally selecting rigid body mark points and optical motion capture system | |
| CN111709985A (en) | Underwater target ranging method based on binocular vision | |
| CN109341559A (en) | A kind of aero-engine casing deformation measurement method based on Binocular Vision Principle | |
| CN113706587A (en) | Rapid point cloud registration method, device and equipment based on space grid division | |
| CN111882596B (en) | Three-dimensional imaging method and device for structured light module, electronic equipment and storage medium | |
| CN108460797B (en) | Method and device for calculating relative pose of depth camera and height of scene plane | |
| CN116952132A (en) | Partition calibration method and system for multi-vision measurement | |
| CN113298885A (en) | Binocular calibration method and device, equipment and storage medium | |
| CN104322057A (en) | Pattern processing apparatus, pattern processing method, and pattern processing program | |
| CN114998397B (en) | Multi-view satellite image stereopair optimization selection method | |
| CN115719387A (en) | 3D camera calibration method, point cloud image acquisition method and camera calibration system | |
| CN113589263B (en) | Method and system for jointly calibrating multiple homologous sensors | |
| Toschi et al. | Improving automated 3D reconstruction methods via vision metrology | |
| CN115810052A (en) | Camera calibration method and device, electronic equipment and storage medium | |
| CN115018922A (en) | Distortion parameter calibration method, electronic device and computer readable storage medium | |
| CN107783310A (en) | A kind of scaling method and device of post lens imaging system | |
| CN115442591B (en) | Camera quality testing method, system, electronic device and storage medium | |
| CN118097039B (en) | Surface structured light three-dimensional point cloud reconstruction method and system based on discrete projection light |
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 |