CN108898636B - Camera one-dimensional calibration method based on improved PSO - Google Patents

Camera one-dimensional calibration method based on improved PSO Download PDF

Info

Publication number
CN108898636B
CN108898636B CN201810583687.4A CN201810583687A CN108898636B CN 108898636 B CN108898636 B CN 108898636B CN 201810583687 A CN201810583687 A CN 201810583687A CN 108898636 B CN108898636 B CN 108898636B
Authority
CN
China
Prior art keywords
calibration
dimensional
camera
value
pso
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
CN201810583687.4A
Other languages
Chinese (zh)
Other versions
CN108898636A (en
Inventor
吴丽君
张宇煌
陈志聪
程树英
林丽霞
林培杰
郑茜颖
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.)
Fuzhou University
Original Assignee
Fuzhou University
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 Fuzhou University filed Critical Fuzhou University
Priority to CN201810583687.4A priority Critical patent/CN108898636B/en
Publication of CN108898636A publication Critical patent/CN108898636A/en
Application granted granted Critical
Publication of CN108898636B publication Critical patent/CN108898636B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Length Measuring Devices By Optical Means (AREA)
  • Studio Devices (AREA)

Abstract

The invention relates to a camera one-dimensional calibration method based on improved PSO (particle swarm optimization), which is characterized in that the results of multiple times of camera calibration are clustered according to a clustering algorithm, and then the average value of target categories is taken as the initial value of subsequent PSO optimization; then, the obtained initial value is adjusted by utilizing the prior information; and finally, optimizing the adjusted initial value by applying a standard PSO (particle swarm optimization), thereby obtaining more accurate camera parameters. The invention can acquire the camera precision more accurately and efficiently, thereby improving the precision of vision measurement.

Description

Camera one-dimensional calibration method based on improved PSO
Technical Field
The invention relates to the field of computer vision, in particular to a camera one-dimensional calibration method based on improved PSO.
Background
The calibration of the camera makes it possible to acquire three-dimensional information from two-dimensional images. The existing calibration techniques can be classified into different categories according to different classification criteria. Calibration algorithms can be generally classified into three-dimensional calibration, two-dimensional calibration, one-dimensional calibration, and zero-dimensional calibration (self-calibration) according to the dimensions of the calibration target. One-dimensional calibration has attracted much attention since the introduction of Zhangyiyouyou. The one-dimensional calibration makes use of the actual distance of the known calibration object, and the invariance of the calibration object under projective transformation by combining the internal relation of the mark points, so that the solution of the camera parameters becomes possible. Compared with other calibration algorithms: the calibration precision of the one-dimensional calibration algorithm is superior to zero-dimensional calibration; the operation complexity and the availability of calibration materials are obviously superior to three-dimensional calibration and two-dimensional calibration.
Because the image is difficult to avoid the pollution of noise in the acquisition process, the camera parameters obtained by directly utilizing the calibration algorithm to solve often have larger errors. To solve this problem, non-linear optimization, such as LM optimization, is usually added after linear solution, and the calibration accuracy is improved by reducing back-projection errors. However, such a nonlinear optimization method depends heavily on an initial value and is easy to fall into a local minimum value; meanwhile, the one-sided pursuit of mathematical optimization may make the optimized result meaningless, i.e., over-optimized. In recent years, improved particle swarm optimization algorithms are continuously applied to the field of camera calibration, and although the algorithms increase the calibration accuracy to a certain extent, the existing PSO optimization algorithm depends heavily on initial values and tends to be over-optimized.
Disclosure of Invention
In view of this, the present invention provides a camera one-dimensional calibration method based on an improved PSO, and an algorithm can acquire the camera precision more accurately and efficiently, so as to improve the precision of vision measurement.
The invention is realized by adopting the following scheme: a camera one-dimensional calibration method based on improved PSO specifically comprises the following steps:
step S1: performing one-dimensional calibration on the camera, and obtaining a matrix about parameters of the camera according to the results obtained by multiple times of camera calibration;
step S2: clustering the camera parameter matrix obtained in the step S1 to obtain a parameter type with a better calibration result, and averaging the parameter type to obtain a preliminary initial value;
step S3: adjusting the initial value obtained in the step S2 by using prior information to obtain an adjusted initial value;
step S4: and (5) taking the adjusted initial value obtained in the step (S3) as an input, and optimizing the adjusted initial value by adopting a standard PSO (particle swarm optimization) to obtain a camera parameter.
According to the invention, a reasonable initial value is obtained through a self-clustering algorithm, and then related parameters are limited through prior information, so that the PSO optimization is ensured to develop towards a correct direction.
Further, step S1 specifically includes the following steps:
step S11: manufacturing a one-dimensional calibration target;
step S12: acquiring a calibration object image;
step S13: grouping the calibration object images obtained in the step S2, and extracting the image coordinates of the markers in each frame of image for calibration;
step S14: and acquiring multiple calibration results, and forming a calibration parameter matrix by discharging each calibration result according to a row.
Preferably, the core of the cluster-based data preprocessing is to utilize the redundancy of the data and further mine the data to obtain an initial value better than the mean value. The extreme value of the calibration result is accidental; in contrast, most of the calibration results are gathered around a certain calibration result. Thus, the different calibration results are not considered to be points distributed within a high dimensional sphere, and the desired target class can be obtained by finding the appropriate radius and center of sphere. The input of the whole algorithm is a parameter matrix of multiple calibration results, and each calibration result occupies one row.
Thus, the specific clustering algorithm of step S2 is: the method comprises the following steps:
step S21: sequencing the multiple calibration results to find out the maximum value P of the calibration resultsmax、PminAnd remember PmaxAnd PminThe Euclidean distance between the two is d;
step S22: determining the number of points contained in the target category;
step S23: initializing a clustering algorithm, setting different calibration results and finally distributing the calibration results in a high-dimensional ball, and determining an initial center, an initial radius and a step length of the high-dimensional ball;
step S24: calculating the Euclidean distance between each calibration point and the center of the sphere, and judging whether the point is in the high-dimensional sphere or not;
step S25: accumulating the index points satisfying the step S24, and determining whether the accumulated points satisfy the points determined in the step S22; if not, adjusting the radius of the high-dimensional sphere or changing the center position of the sphere, and returning to the step S24; otherwise, go to step S26;
step S26: the object classes that lie within the high-dimensional sphere are averaged.
Further, step S22 is specifically: according to the calibration knotMaximum value of fruit Pmax、PminThe Euclidean distance d between the target classes is calculated by adopting the following formula:
Figure BDA0001688985750000031
further, in step S23, the initial sphere center is set as the first row element of the parameter matrix, the initial radius is set as d/10, and the step size is d/100.
Further, in step S25, the radius of the high-dimensional sphere is adjusted in units of one step; and transforming the position of the sphere center to traverse the row elements of the whole matrix according to a browsing method.
By adopting the steps, a better PSO optimization initial value can be obtained. Although a good PSO optimization initial value can be obtained by using the algorithm, the good initial value cannot guarantee that the optimization result is converged towards the true value of the camera parameter. Since the PSO optimization algorithm simply pursues the most significant value of the objective function, it is necessary to add certain limiting factors to ensure the physical significance of the optimization result. Therefore, the invention also provides a PSO optimization algorithm based on prior information.
Further, the step S3 is specifically: correcting the initial value of the principal point coordinate by using the prior physical information, and generating a search range of 8-dimensional particles according to the mean square error of a parameter matrix calibrated by a camera for multiple times; wherein, the existing prior physical information comprises: the camera principal point is near the center of the imaging sensor.
Specifically, the existing a priori physical information further includes: typically the camera principal point is about half the size of the original picture of the camera. Specifically, therefore, in step S3, the corresponding parameter in the clustering result is replaced with half the size of the picture. And the mean square error used to generate the particles is derived from the statistics of the parameter matrix.
Further, step S4 specifically includes the following steps:
step S41: initializing a PSO;
step S42: the fitness value for each particle was calculated according to the following formula:
Figure BDA0001688985750000041
in the formula, xpiAnd ypiRespectively representing the abscissa, x, of the reprojected pointsiAnd yiRespectively representing the real horizontal and vertical coordinates of the projection points, and N representing the number of the populations;
step S43: updating the individual best fit value if the current fit value is better than the best fit value of the particle
Figure BDA0001688985750000042
If the current population adaptation value is superior to the historical optimal adaptation degree of the population, updating the optimal adaptation value of the population
Figure BDA0001688985750000043
Step S44: and adding one to the iteration number, judging whether the iteration number reaches an iteration number or an end threshold value, if so, ending the optimization, and otherwise, returning to the step S43.
Further, the step S41 is specifically: defining the number N of the population, the number G of iterations, the inertia weight w and the acceleration constant c1、c2And an end threshold.
Further, the number of population N is 30, the number of iterations G is 500, the inertia weight w is 0.8, and the acceleration constant c is set to be equal to1＝1.2，c2＝2.3。
Compared with the prior art, the invention has the following beneficial effects:
1. compared with the contingency of a single experiment, the accuracy can be improved by depending on the redundancy of data in multiple experiments.
2. Compared with the method for averaging multiple calibration results, the optimization algorithm provided by the invention has higher precision and smaller error, and can still ensure certain precision under the condition of large noise.
3. Compared with the existing improved PSO algorithm, the initial value acquisition and over optimization problems are optimized to a certain degree, so that the optimization result is more accurate and more efficient.
Drawings
FIG. 1 is a schematic diagram of the method of the embodiment of the present invention.
Fig. 2 is a one-dimensional calibration schematic diagram of a camera according to an embodiment of the present invention.
Fig. 3 is a schematic diagram of parameter errors at different noise levels during simulation according to an embodiment of the present invention.
Fig. 4 is a schematic diagram of image data acquired by two-dimensional calibration according to an embodiment of the present invention.
Fig. 5 is image data acquired by one-dimensional calibration according to an embodiment of the present invention.
Detailed Description
The invention is further explained below with reference to the drawings and the embodiments.
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
As shown in fig. 1, the present embodiment provides a camera one-dimensional calibration method based on an improved PSO, where a camera to be calibrated is GT1910C of allid, and an FC f with a lens of compotar is 25mm. Intel Core (TM) i5-4430S, MATLAB2016 b.
The method specifically comprises the following steps:
step S1: performing one-dimensional calibration on the camera, and obtaining a matrix about parameters of the camera according to the results obtained by multiple times of camera calibration;
step S2: clustering the camera parameter matrix obtained in the step S1 to obtain a parameter type with a better calibration result, and averaging the parameter type to obtain a preliminary initial value;
step S3: adjusting the initial value obtained in the step S2 by using prior information to obtain an adjusted initial value;
step S4: and (5) taking the adjusted initial value obtained in the step (S3) as an input, and optimizing the adjusted initial value by adopting a standard PSO (particle swarm optimization) to obtain a camera parameter.
According to the invention, a reasonable initial value is obtained through a self-clustering algorithm, and then related parameters are limited through prior information, so that the PSO optimization is ensured to develop towards a correct direction.
In this embodiment, as shown in fig. 2, fig. 2 is a one-dimensional calibration schematic diagram of a camera, and step S1 specifically includes the following steps:
step S11: manufacturing a one-dimensional calibration target: three markers are fixed on the thin stick respectively. The interval between every two markers is 13cm, namely the effective length of the thin stick is 26 cm;
step S12: obtaining a calibration object image: three-dimensionally rotating the calibration target around the marker of the thin stick end point, and shooting the motion process by using a camera to be calibrated to obtain a thin stick motion image with about 400 frames, wherein 4 pieces of the motion image are shown in fig. 4, wherein fig. 4(a) is a 25 th frame, fig. 4(b) is a 150 th frame, fig. 4(c) is a 275 th frame, and fig. 4(d) is a 315 th frame;
step S13: the calibration object images obtained in step S2 are grouped, and the image coordinates of the markers in each frame of image are extracted for calibration: dividing 400 frames of images into 20 groups, and extracting image coordinates of markers in each frame of image for calibration;
step S14: and acquiring multiple calibration results, and forming a calibration parameter matrix by discharging each calibration result according to a row.
Preferably, the core of the cluster-based data preprocessing is to utilize the redundancy of the data and further mine the data to obtain an initial value better than the mean value. The extreme value of the calibration result is accidental; in contrast, most of the calibration results are gathered around a certain calibration result. Thus, the different calibration results are not considered to be points distributed within a high dimensional sphere, and the desired target class can be obtained by finding the appropriate radius and center of sphere. The input of the whole algorithm is a parameter matrix of multiple calibration results, and each calibration result occupies one row.
Thus, in this embodiment, the specific clustering algorithm of step S2 is: the method comprises the following steps:
step S21: sequencing the multiple calibration results to find out the maximum value P of the calibration resultsmax、PminAnd remember PmaxAnd PminThe Euclidean distance between the two is d;
step S22: determining the number of points contained in the target category;
step S23: initializing a clustering algorithm, setting different calibration results and finally distributing the calibration results in a high-dimensional ball, and determining an initial center, an initial radius and a step length of the high-dimensional ball;
step S24: calculating the Euclidean distance between each calibration point and the center of the sphere, and judging whether the point is in the high-dimensional sphere or not;
step S25: accumulating the index points satisfying the step S24, and determining whether the accumulated points satisfy the points determined in the step S22; if not, adjusting the radius of the high-dimensional sphere or changing the center position of the sphere, and returning to the step S24; otherwise, go to step S26;
step S26: the object classes that lie within the high-dimensional sphere are averaged.
In this embodiment, step S22 specifically includes: according to the maximum value P of the calibration resultmax、PminThe Euclidean distance d between the target classes is calculated by adopting the following formula:
Figure BDA0001688985750000071
in this embodiment, in step S23, the initial center of sphere is set as the first row element of the parameter matrix, the initial radius is set as d/10, and the step size is d/100.
In this embodiment, in step S25, the radius of the high-dimensional sphere is adjusted in units of one step; and transforming the position of the sphere center to traverse the row elements of the whole matrix according to a browsing method.
By adopting the steps, a better PSO optimization initial value can be obtained. Although a good PSO optimization initial value can be obtained by using the algorithm, the good initial value cannot guarantee that the optimization result is converged towards the true value of the camera parameter. Since the PSO optimization algorithm simply pursues the most significant value of the objective function, it is necessary to add certain limiting factors to ensure the physical significance of the optimization result. Therefore, the invention also provides a PSO optimization algorithm based on prior information.
In this embodiment, the step S3 specifically includes: correcting the initial value of the principal point coordinate by using the prior physical information, and generating a search range of 8-dimensional particles according to the mean square error of a parameter matrix calibrated by a camera for multiple times to obtain an adjusted initial value; wherein, the existing prior physical information comprises: the camera principal point is near the center of the imaging sensor.
Specifically, in this embodiment, the existing a priori physical information further includes: typically the camera principal point is about half the size of the original picture of the camera. Specifically, therefore, in step S3, the corresponding parameter in the clustering result is replaced with half the size of the picture. And the mean square error used to generate the particles is derived from the statistics of the parameter matrix.
In this embodiment, step S4 specifically includes the following steps:
step S41: initializing a PSO;
step S42: the fitness value for each particle was calculated according to the following formula:
Figure BDA0001688985750000081
in the formula, xpiAnd ypiRespectively representing the abscissa, x, of the reprojected pointsiAnd yiRespectively representing the real horizontal and vertical coordinates of the projection points, and N representing the number of the populations;
step S43: updating the individual best fit value if the current fit value is better than the best fit value of the particle
Figure BDA0001688985750000082
If the current population adaptation value is superior to the historical optimal adaptation degree of the population, updating the optimal adaptation value of the population
Figure BDA0001688985750000083
Step S44: and adding one to the iteration number, judging whether the iteration number reaches an iteration number or an end threshold value, if so, ending the optimization, and otherwise, returning to the step S43.
In this embodiment, the step S41 specifically includes: defining the number N of the population, the number G of iterations, the inertia weight w and the acceleration constant c1、c2And an end threshold.
In the present embodiment, the number of clusters N is 30, the number of iterations G is 500, the inertia weight w is 0.8, and the acceleration constant c is set to be equal to1＝1.2，c2＝2.3。
Specifically, in the present embodiment, in order to verify the correctness of the above algorithm, assume that the one-dimensional calibration camera parameters P ═ 1000,0,320,240,0,45,150, and assume that the noise levels σ are 0.1, 0.2, 0.3, 0.4, 0.5, respectively; each noise level was scaled 20 times. The results of the computer simulation are shown in fig. 3. Fig. 3(a) is an error graph of direct averaging of multiple calibration results, and fig. 3(b) is an error graph of the method of the present embodiment. In contrast, it is clear that the optimization results using this embodiment are higher in accuracy than the results of direct averaging.
The relevant calibration results of this embodiment are shown in table 1, and for comparison, the results of two-dimensional zhang calibration are used as standard values in this embodiment. In the experiment, a total of 10 pictures of the board movement are collected for calibration, two of which are shown in fig. 5, wherein fig. 5(a) is the first picture and fig. 5(b) is the seventh picture. Experiments prove that the method provided by the embodiment is effective, high-precision and realizable.
TABLE 1 derived camera part parameters for different methods
Figure BDA0001688985750000091
The above description is only a preferred embodiment of the present invention, and all equivalent changes and modifications made in accordance with the claims of the present invention should be covered by the present invention.

Claims (8)

1. A camera one-dimensional calibration method based on improved PSO is characterized in that:
step S1: performing one-dimensional calibration on the camera, and obtaining a matrix related to camera parameters according to the results obtained by multiple times of camera calibration;
step S2: clustering the camera parameter matrix obtained in the step S1 to obtain a parameter type with a better calibration result, and averaging the parameter type to obtain a preliminary initial value;
step S3: adjusting the initial value obtained in the step S2 by using prior information to obtain an adjusted initial value;
step S4: taking the adjusted initial value obtained in the step S3 as input, and optimizing the adjusted initial value by adopting a standard PSO (particle swarm optimization) to obtain camera parameters;
step S2 specifically includes: the method comprises the following steps:
step S21: sequencing the multiple calibration results to find out the maximum value P of the calibration resultsmax、PminAnd remember PmaxAnd PminThe Euclidean distance between the two is d;
step S22: determining the number of points contained in the target category;
step S23: initializing a clustering algorithm, setting different calibration results and finally distributing the calibration results in a high-dimensional ball, and determining an initial center, an initial radius and a step length of the high-dimensional ball;
step S24: calculating the Euclidean distance between each calibration point and the center of the sphere, and judging whether the point is in the high-dimensional sphere or not;
step S25: accumulating the index points satisfying the step S24, and determining whether the accumulated points satisfy the points determined in the step S22; if not, adjusting the radius of the high-dimensional sphere or changing the center position of the sphere, and returning to the step S24; otherwise, go to step S26;
step S26: averaging the target categories located in the high-dimensional ball;
the step S3 specifically includes: correcting the initial value of the principal point coordinate by using the prior physical information, and generating a search range of 8-dimensional particles according to the mean square error of a parameter matrix calibrated by a camera for multiple times; wherein, the existing prior physical information comprises: the camera principal point is near the center of the imaging sensor.
2. The one-dimensional calibration method for the camera based on the improved PSO as claimed in claim 1, wherein: step S1 specifically includes the following steps:
step S11: manufacturing a one-dimensional calibration target;
step S12: acquiring a calibration object image;
step S13: grouping the calibration object images obtained in the step S12, and extracting the image coordinates of the markers in each frame of image for calibration;
step S14: and acquiring multiple calibration results, and forming a calibration parameter matrix by discharging each calibration result according to a row.
3. The one-dimensional calibration method for the camera based on the improved PSO as claimed in claim 2, wherein: step S22 specifically includes: according to the maximum value P of the calibration resultmax、PminThe Euclidean distance d between the target classes is calculated by adopting the following formula:
Figure FDA0003385975420000021
4. the one-dimensional calibration method for the camera based on the improved PSO as claimed in claim 2, wherein: in step S23, the initial sphere center is set as the first row element of the parameter matrix, the initial radius is set as d/10, and the step size is d/100.
5. The one-dimensional calibration method for the camera based on the improved PSO as claimed in claim 2, wherein: in step S25, the radius of the high-dimensional sphere is adjusted in units of one step; and transforming the position of the sphere center to traverse the row elements of the whole matrix according to a browsing method.
6. The one-dimensional calibration method for the camera based on the improved PSO as claimed in claim 1, wherein: step S4 specifically includes the following steps:
step S41: initializing a PSO;
step S42: the fitness value for each particle was calculated according to the following formula:
Figure FDA0003385975420000022
in the formula, xpiAnd ypiRespectively representing the abscissa, x, of the reprojected pointsiAnd yiRespectively representing the real horizontal and vertical coordinates of the projection points, and N representing the number of the populations;
step S43: updating the individual best fit value if the current fit value is better than the best fit value of the particle
Figure FDA0003385975420000023
If the current population adaptation value is superior to the historical optimal adaptation degree of the population, updating the optimal adaptation value of the population
Figure FDA0003385975420000024
Step S44: and adding one to the iteration number, judging whether the iteration number reaches an iteration number or an end threshold value, if so, ending the optimization, and otherwise, returning to the step S43.
7. The one-dimensional calibration method for the camera based on the improved PSO as claimed in claim 6, wherein: the step S41 specifically includes: defining the number N of the population, the number G of iterations, the inertia weight w and the acceleration constant c1、c2And an end threshold.
8. The one-dimensional calibration method for camera based on improved PSO as claimed in claim 7, wherein: the number of groups N is 30, the number of iterations G is 500, the inertia weight w is 0.8, and the acceleration constant c is set to be equal to1＝1.2，c2＝2.3。
CN201810583687.4A 2018-06-08 2018-06-08 Camera one-dimensional calibration method based on improved PSO Active CN108898636B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810583687.4A CN108898636B (en) 2018-06-08 2018-06-08 Camera one-dimensional calibration method based on improved PSO

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810583687.4A CN108898636B (en) 2018-06-08 2018-06-08 Camera one-dimensional calibration method based on improved PSO

Publications (2)

Publication Number Publication Date
CN108898636A CN108898636A (en) 2018-11-27
CN108898636B true CN108898636B (en) 2022-03-29

Family

ID=64344402

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810583687.4A Active CN108898636B (en) 2018-06-08 2018-06-08 Camera one-dimensional calibration method based on improved PSO

Country Status (1)

Country Link
CN (1) CN108898636B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110853104B (en) * 2020-01-15 2020-05-05 广东博智林机器人有限公司 Calibration plate, machine vision calibration device and method

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101839692B (en) * 2010-05-27 2012-09-05 西安交通大学 Method for measuring three-dimensional position and stance of object with single camera
CN102376093B (en) * 2011-10-13 2014-04-23 清华大学 Calibration method of camera
JP5886594B2 (en) * 2011-10-27 2016-03-16 セコム株式会社 Camera system
US20140257767A1 (en) * 2013-03-09 2014-09-11 Bigwood Technology, Inc. PSO-Guided Trust-Tech Methods for Global Unconstrained Optimization
CN105631844A (en) * 2014-10-28 2016-06-01 上海大学 Image camera calibration method
CN107153837A (en) * 2017-04-14 2017-09-12 中国科学技术大学苏州研究院 Depth combination K means and PSO clustering method

Also Published As

Publication number Publication date
CN108898636A (en) 2018-11-27

Similar Documents

Publication Publication Date Title
WO2022002150A1 (en) Method and device for constructing visual point cloud map
CN111780763B (en) Visual positioning method and device based on visual map
CN107886529B (en) Point cloud registration method for three-dimensional reconstruction
CN106845621B (en) Dense population number method of counting and system based on depth convolutional neural networks
CN112328715B (en) Visual positioning method, training method of related model, related device and equipment
CN110310310B (en) Improved method for aerial image registration
CN108038435B (en) Feature extraction and target tracking method based on convolutional neural network
CN110930411B (en) Human body segmentation method and system based on depth camera
CN109544603B (en) Target tracking method based on deep migration learning
CN109325510B (en) Image feature point matching method based on grid statistics
CN110555377B (en) Pedestrian detection and tracking method based on fish eye camera overlooking shooting
CN105930790A (en) Human body behavior recognition method based on kernel sparse coding
CN105141872B (en) The method of video image is handled when a kind of contracting
CN110443279B (en) Unmanned aerial vehicle image vehicle detection method based on lightweight neural network
CN113393524B (en) Target pose estimation method combining deep learning and contour point cloud reconstruction
WO2020107312A1 (en) Rigid body configuration method and optical motion capturing method
CN112529945B (en) Multi-view three-dimensional ISAR scattering point set registration method
CN107704867A (en) Based on the image characteristic point error hiding elimination method for weighing the factor in a kind of vision positioning
CN111310821A (en) Multi-view feature fusion method, system, computer device and storage medium
CN108898636B (en) Camera one-dimensional calibration method based on improved PSO
CN109857895B (en) Stereo vision retrieval method and system based on multi-loop view convolutional neural network
CN108447038B (en) Grid denoising method based on non-local total variation operator
CN111161348A (en) Monocular camera-based object pose estimation method, device and equipment
CN116894876A (en) 6-DOF positioning method based on real-time image
CN115246125A (en) Robot vision servo control method and system based on hybrid feedback

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