CN108898636A - A kind of camera one-dimension calibration method based on improvement PSO - Google Patents
A kind of camera one-dimension calibration method based on improvement PSO Download PDFInfo
- Publication number
- CN108898636A CN108898636A CN201810583687.4A CN201810583687A CN108898636A CN 108898636 A CN108898636 A CN 108898636A CN 201810583687 A CN201810583687 A CN 201810583687A CN 108898636 A CN108898636 A CN 108898636A
- Authority
- CN
- China
- Prior art keywords
- camera
- dimension
- calibration
- 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.)
- Granted
Links
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)
- Studio Devices (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The present invention relates to a kind of based on the camera one-dimension calibration method for improving PSO, is clustered to the result of multiple camera calibration according to clustering algorithm, then to the initial value that target category takes mean value to optimize as subsequent PSO;It adjusts followed by prior information to obtained initial value;Finally initial value adjusted is done optimize using standard PSO, to obtain more accurately camera parameter.The present invention can more accurate, efficient acquisition camera precision, to promote the precision of vision measurement.
Description
Technical field
It is especially a kind of based on the camera one-dimension calibration method for improving PSO the present invention relates to computer vision field.
Background technique
The calibration of camera becomes possibility so that obtaining three-dimensional information from two dimensional image.Existing calibration technique is according to difference
Classification standard can be divided into different classifications.Usually calibration algorithm can be divided into three according to the dimension difference of calibration target
Tie up calibration, two-dimensional calibrations, one-dimension calibration and zero dimension calibration (self-calibration).One-dimension calibration is after Zhang Zhengyou proposition just by pass
Note.One-dimension calibration utilizes the actual range of known calibration object, in conjunction with the inner link of index point on calibration object, using it in projection
Invariance under transformation makes it possible the solution of camera parameter.Compared to other calibration algorithms:The mark of one-dimension calibration algorithm
Determine precision to demarcate better than zero dimension;The triviality of operation and the ready availability of calibration material are also significantly better than three-dimensional scaling and two dimension mark
It is fixed.
Due to image in collection process inevitably by the pollution of noise, the camera that is directly solved using calibration algorithm
Often there is biggish error in parameter.In order to solve this problem, nonlinear optimization is usually added after linear solution, as LM is excellent
Change, improves stated accuracy by reducing back projection's error.But this kind of nonlinear optimization method heavy dependence initial value, it is easily trapped into
Local minimum;Meanwhile the optimal result that may make optimization in mathematics that covets loses meaning i.e. excessively optimization.In recent years
Some modified particle swarm optiziations are constantly applied to camera calibration field, although these algorithms increase mark to a certain extent
Fixed precision, but existing PSO optimization algorithm heavy dependence initial value, and there is the tendency excessively optimized.
Summary of the invention
In view of this, algorithm can the purpose of the present invention is to propose to a kind of camera one-dimension calibration method based on improvement PSO
More accurate, efficient acquisition camera precision, to promote the precision of vision measurement.
The present invention is realized using following scheme:A kind of camera one-dimension calibration method based on improvement PSO, specifically includes following
Step:
Step S1:Camera one-dimension calibration is carried out, according to the acquisition of multiple camera calibration as a result, obtaining about progress camera ginseng
Several matrixes;
Step S2:The camera parameter matrix that step S1 is obtained is clustered, the preferable clock rate of calibration result is obtained
And it is averaged, obtain preliminary initial value;
Step S3:It is adjusted using prior information to the preliminary initial value that step S2 is obtained, the initial value after being adjusted;
Step S4:The initial value adjusted that step S3 is obtained is as input, using standard PSO to adjusted initial
Value, which is done, to be optimized, and camera parameter is obtained.
The present invention first passes through from clustering algorithm and obtains relatively reasonable ground initial value, then by prior information to relevant parameter into
Row limits, to guarantee that PSO optimization develops towards correct direction.
Further, step S1 specifically includes following steps:
Step S11:Make one-dimension calibration target;
Step S12:Obtain reference object image;
Step S13:The step S2 reference object image obtained is grouped, and extracts the image of marker in every frame image
Coordinate, for demarcating;
Step S14:Multiple calibration result is obtained, each calibration result forms calibrating parameters matrix by row discharge.
Preferably, the core of the data prediction based on cluster is the redundancy using data, data are carried out further
It excavates to obtain than taking the better initial value of mean value.There is extremum with contingency in calibration result;In comparison, big portion
Divide calibration result that can be gathered near some calibration result.Therefore, different calibration results might as well be regarded as and is distributed across one
The intracorporal point of higher-dimension ball can be obtained by required target category by looking for suitable radius and the centre of sphere.Entirely algorithm is defeated
Enter for the parameter matrix of multiple calibration result, each calibration result occupies a line.
The specific clustering algorithm of step S2 is as a result,:Include the following steps:
Step S21:Multiple calibration result is ranked up, the most value P of calibration result is found outmax、Pmin, and remember PmaxWith Pmin
Between Euclidean distance be d;
Step S22:Determine the points that target category includes;
Step S23:Clustering algorithm is initialized, if different calibration results finally will be distributed in a higher-dimension ball, is determined
The initial centre of sphere of higher-dimension ball, initial radium and step-length;
Step S24:Euclidean distance of each calibration point apart from the centre of sphere is calculated, and judges the point whether in the higher-dimension ball
It is interior;
Step S25:The cumulative calibration point for meeting step S24, and judge whether the points after adding up meet step S22 and determine
Points;If not satisfied, then adjusting the radius or transformation sphere center position of the higher-dimension ball, and return step S24;Otherwise, enter
Step S26;
Step S26:Mean value is taken to the target category being located in higher-dimension ball.
Further, step S22 is specially:According to the most value P of calibration resultmax、PminBetween Euclidean distance d, use
Following formula calculates the point number that target category should include:
Further, in step S23, the initial centre of sphere is set as the first row element of parameter matrix, and initial radium is set as d/
10, step-length d/100.
Further, in step S25, the radius of the higher-dimension ball is adjusted as unit of a step-length;Convert sphere center position
The row element of entire matrix is traversed according to the method for browsing.
Preferable PSO optimized initial value can be obtained using above-mentioned steps.Using above-mentioned algorithm although it is available compared with
Good PSO optimizes initial value, but preferable initial value does not ensure that the result of optimization is centainly received towards camera parameter true value
It holds back.Due to the most value of the simple function of pursuit of the objective of PSO optimization algorithm, it is therefore necessary to it is excellent to guarantee that certain limiting factor be added
Change the physical significance of result.For this purpose, the present invention proposes PSO optimization algorithm based on prior information simultaneously.
Further, the step S3 is specially:Using existing priori physical message to the initial value of principal point coordinate into
The mean square deviation of row amendment and the parameter matrix according to multiple camera calibration generates the search range of the particle of 8 dimensions;Wherein, have
Priori physical message include:Immediate vicinity of the principal point for camera in imaging sensor.
Particularly, the existing priori physical message further includes:Usual principal point for camera is about camera original image size
Half.Therefore specifically, in step s3, with corresponding parameter in the half substitution cluster result of dimension of picture.And generate grain
Statistical result of the mean square deviation used in son from parameter matrix.
Further, step S4 specifically includes following steps:
Step S41:Initialize PSO;
Step S42:The adaptive value of each particle is calculated according to the following formula:
In formula, xpiWith ypiRespectively indicate the transverse and longitudinal coordinate of re-projection point, xiWith yiRespectively indicate the true transverse and longitudinal of subpoint
Coordinate, N indicate population number;
Step S43:If current adaptive value is better than the optimal adaptation value of particle, more new individual optimal adaptation value
If the current Population adaptation value of fruit is better than the history optimal adaptation degree of population, Population Regeneration optimal adaptation value
Step S44:The number of iterations adds one, and judges whether to reach the number of iterations or terminate threshold value, if then terminating excellent
Change, otherwise, return step S43.
Further, the step S41 is specially:It is normal to specify population number N, the number of iterations G, inertia weight w, acceleration
Number c1、c2With end threshold value.
Further, population number N=30, the number of iterations G=500, inertia weight w=0.8, acceleration constant c1=
1.2, c2=2.3.
Compared with prior art, the invention has the following beneficial effects:
1, the contingency of the invention compared to single experiment, the redundancy that many experiments can rely on data improve precision.
2, mean value being taken compared to multiple calibration result, optimization algorithm precision proposed by the present invention is higher, and error is smaller, and
Again under big noise, it can still guarantee certain precision.
3, it is all carried out on compared to existing improvement PSO algorithm, the acquisition of initial value of the present invention and excessive optimization the problems such as
A degree of optimization, therefore optimum results are more accurate, it is more efficient.
Detailed description of the invention
Fig. 1 is the Method And Principle schematic diagram of the embodiment of the present invention.
Fig. 2 is the camera one-dimension calibration schematic diagram of the embodiment of the present invention.
Parameter error schematic diagram when Fig. 3 is the emulation of the embodiment of the present invention under difference noise level.
Fig. 4 is the image data schematic diagram that the two-dimensional calibrations of the embodiment of the present invention acquire.
Fig. 5 is the image data that the one-dimension calibration of the embodiment of the present invention acquires.
Specific embodiment
The present invention will be further described with reference to the accompanying drawings and embodiments.
It is noted that following detailed description is all illustrative, it is intended to provide further instruction to the application.Unless another
It indicates, all technical and scientific terms used herein has usual with the application person of an ordinary skill in the technical field
The identical meanings of understanding.
It should be noted that term used herein above is merely to describe specific embodiment, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singular
Also it is intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or their combination.
As shown in Figure 1, present embodiments providing a kind of based on the camera one-dimension calibration method for improving PSO, camera to be calibrated
For the GT1910C of ALLIED company, the FC f=25mm. data processing platform (DPP) that camera lens is COMPUTAR is:Intel Core
(TM) i5-4430S, MATLAB2016b.
Specifically include following steps:
Step S1:Camera one-dimension calibration is carried out, according to the acquisition of multiple camera calibration as a result, obtaining about progress camera ginseng
Several matrixes;
Step S2:The camera parameter matrix that step S1 is obtained is clustered, the preferable clock rate of calibration result is obtained
And it is averaged, obtain preliminary initial value;
Step S3:It is adjusted using prior information to the preliminary initial value that step S2 is obtained, the initial value after being adjusted;
Step S4:The initial value adjusted that step S3 is obtained is as input, using standard PSO to adjusted initial
Value, which is done, to be optimized, and camera parameter is obtained.
The present invention first passes through from clustering algorithm and obtains relatively reasonable ground initial value, then by prior information to relevant parameter into
Row limits, to guarantee that PSO optimization develops towards correct direction.
In the present embodiment, as shown in Fig. 2, Fig. 2 is camera one-dimension calibration schematic diagram, step S1 specifically includes following step
Suddenly:
Step S11:Make one-dimension calibration target:Three markers are taken, are separately fixed on thin rod.And marker two-by-two it
Between between be divided into 13cm, i.e., thin rod effective length be 26cm;
Step S12:Obtain reference object image:Marker of the target around thin rod endpoint will be demarcated and carry out three-dimensional rotation, be used in combination
Camera to be calibrated shoots motion process, obtains the thin rod moving image of about 400 frames, wherein 4 as shown in figure 4, wherein Fig. 4
It (a) is the 25th frame, Fig. 4 (b) is the 150th frame, and Fig. 4 (c) is the 275th frame, and Fig. 4 (d) is the 315th frame;
Step S13:The step S2 reference object image obtained is grouped, and extracts the image of marker in every frame image
Coordinate, for demarcating:400 frame images are divided into 20 groups, and extract the image coordinate of marker in every frame image, for demarcating;
Step S14:Multiple calibration result is obtained, each calibration result forms calibrating parameters matrix by row discharge.
Preferably, the core of the data prediction based on cluster is the redundancy using data, data are carried out further
It excavates to obtain than taking the better initial value of mean value.There is extremum with contingency in calibration result;In comparison, big portion
Divide calibration result that can be gathered near some calibration result.Therefore, different calibration results might as well be regarded as and is distributed across one
The intracorporal point of higher-dimension ball can be obtained by required target category by looking for suitable radius and the centre of sphere.Entirely algorithm is defeated
Enter for the parameter matrix of multiple calibration result, each calibration result occupies a line.
As a result, in the present embodiment, the specific clustering algorithm of step S2 is:Include the following steps:
Step S21:Multiple calibration result is ranked up, the most value P of calibration result is found outmax、Pmin, and remember PmaxWith Pmin
Between Euclidean distance be d;
Step S22:Determine the points that target category includes;
Step S23:Clustering algorithm is initialized, if different calibration results finally will be distributed in a higher-dimension ball, is determined
The initial centre of sphere of higher-dimension ball, initial radium and step-length;
Step S24:Euclidean distance of each calibration point apart from the centre of sphere is calculated, and judges the point whether in the higher-dimension ball
It is interior;
Step S25:The cumulative calibration point for meeting step S24, and judge whether the points after adding up meet step S22 and determine
Points;If not satisfied, then adjusting the radius or transformation sphere center position of the higher-dimension ball, and return step S24;Otherwise, enter
Step S26;
Step S26:Mean value is taken to the target category being located in higher-dimension ball.
In the present embodiment, step S22 is specially:According to the most value P of calibration resultmax、PminBetween Euclidean distance d,
The point number that target category should include is calculated using following formula:
In the present embodiment, in step S23, the initial centre of sphere is set as the first row element of parameter matrix, initial radium setting
For d/10, step-length d/100.
In the present embodiment, in step S25, the radius of the higher-dimension ball is adjusted as unit of a step-length;Convert the centre of sphere
Position traverses the row element of entire matrix according to the method for browsing.
Preferable PSO optimized initial value can be obtained using above-mentioned steps.Using above-mentioned algorithm although it is available compared with
Good PSO optimizes initial value, but preferable initial value does not ensure that the result of optimization is centainly received towards camera parameter true value
It holds back.Due to the most value of the simple function of pursuit of the objective of PSO optimization algorithm, it is therefore necessary to it is excellent to guarantee that certain limiting factor be added
Change the physical significance of result.For this purpose, the present invention proposes PSO optimization algorithm based on prior information simultaneously.
In the present embodiment, the step S3 is specially:Using existing priori physical message to the initial of principal point coordinate
Value is modified and the mean square deviation of the parameter matrix according to multiple camera calibration, generates the search range of the particle of 8 dimensions, is adjusted
Initial value after whole;Wherein, existing priori physical message includes:Immediate vicinity of the principal point for camera in imaging sensor.
Particularly, in the present embodiment, the existing priori physical message further includes:Usual principal point for camera is about camera
The half of original image size.Therefore specifically, in step s3, with corresponding ginseng in the half substitution cluster result of dimension of picture
Number.And generate statistical result of the mean square deviation used in particle from parameter matrix.
In the present embodiment, step S4 specifically includes following steps:
Step S41:Initialize PSO;
Step S42:The adaptive value of each particle is calculated according to the following formula:
In formula, xpiWith ypiRespectively indicate the transverse and longitudinal coordinate of re-projection point, xiWith yiRespectively indicate the true transverse and longitudinal of subpoint
Coordinate, N indicate population number;
Step S43:If current adaptive value is better than the optimal adaptation value of particle, more new individual optimal adaptation value
If the current Population adaptation value of fruit is better than the history optimal adaptation degree of population, Population Regeneration optimal adaptation value
Step S44:The number of iterations adds one, and judges whether to reach the number of iterations or terminate threshold value, if then terminating excellent
Change, otherwise, return step S43.
In the present embodiment, the step S41 is specially:It specifies population number N, the number of iterations G, inertia weight w, accelerate
Spend constant c1、c2With end threshold value.
In the present embodiment, population number N=30, the number of iterations G=500, inertia weight w=0.8, acceleration constant c1
=1.2, c2=2.3.
Particularly, in the present embodiment, in order to verify the correctness of above-mentioned algorithm, it is assumed that one-dimension calibration camera parameter is P
=[1000,1000,0,320,240,0,45,150], it is assumed that noise level is that σ is 0.1,0.2,0.3,0.4,0.5;Often
A noise level is demarcated 20 times.The result of Computer Simulation is shown in Fig. 3.Fig. 3 (a) is the mistake that mean value is directly taken to multiple calibration result
Difference figure, Fig. 3 (b) are the Error Graph using the present embodiment method.In comparison, hence it is evident that using the optimum results of the present embodiment in essence
Degree aspect is higher than the result for directly taking mean value.
The results are shown in Table 1 for the associated calibration of the present embodiment, and in order to compare, the present embodiment is with the result of Zhang Shi two-dimensional calibrations
As standard value.The present embodiment, which has altogether, in experiment acquires the mobile picture of 10 chessboards for demarcating, wherein two such as Fig. 5 institutes
Show, Fig. 5 (a) is first, and Fig. 5 (b) is the 7th.It is demonstrated experimentally that the method that the present embodiment is proposed is effective, high-precision
, it can be achieved that.
The obtained camera partial parameters of 1 distinct methods of table
The foregoing is merely presently preferred embodiments of the present invention, all equivalent changes done according to scope of the present invention patent with
Modification, is all covered by the present invention.
Claims (10)
1. a kind of based on the camera one-dimension calibration method for improving PSO, it is characterised in that:
Step S1:Camera one-dimension calibration is carried out, according to the acquisition of multiple camera calibration as a result, obtaining about progress camera parameter
Matrix;
Step S2:The camera parameter matrix that step S1 is obtained is clustered, the preferable clock rate of calibration result and right is obtained
It is averaged, and obtains preliminary initial value;
Step S3:It is adjusted using prior information to the preliminary initial value that step S2 is obtained, the initial value after being adjusted;
Step S4:The initial value adjusted that step S3 is obtained does initial value adjusted using standard PSO as input
Optimization, obtains camera parameter.
2. according to claim 1 a kind of based on the camera one-dimension calibration method for improving PSO, it is characterised in that:Step S1
Specifically include following steps:
Step S11:Make one-dimension calibration target;
Step S12:Obtain reference object image;
Step S13:The step S2 reference object image obtained is grouped, and the image for extracting marker in every frame image is sat
Mark, for demarcating;
Step S14:Multiple calibration result is obtained, each calibration result forms calibrating parameters matrix by row discharge.
3. according to claim 1 a kind of based on the camera one-dimension calibration method for improving PSO, it is characterised in that:Step S2
Specially:Include the following steps:
Step S21:Multiple calibration result is ranked up, the most value P of calibration result is found outmax、Pmin, and remember PmaxWith PminBetween
Euclidean distance be d;
Step S22:Determine the points that target category includes;
Step S23:It initializes clustering algorithm and determines higher-dimension if different calibration results finally will be distributed in a higher-dimension ball
The initial centre of sphere of ball, initial radium and step-length;
Step S24:Euclidean distance of each calibration point apart from the centre of sphere is calculated, and judges the point whether in the higher-dimension ball;
Step S25:The cumulative calibration point for meeting step S24, and judge whether the points after adding up meet the point that step S22 is determined
Number;If not satisfied, then adjusting the radius or transformation sphere center position of the higher-dimension ball, and return step S24;Otherwise, it enters step
S26;
Step S26:Mean value is taken to the target category being located in higher-dimension ball.
4. according to claim 3 a kind of based on the camera one-dimension calibration method for improving PSO, it is characterised in that:Step S22
Specially:According to the most value P of calibration resultmax、PminBetween Euclidean distance d, calculating target category using following formula should wrap
The point number contained:
5. according to claim 3 a kind of based on the camera one-dimension calibration method for improving PSO, it is characterised in that:Step S23
In, the initial centre of sphere is set as the first row element of parameter matrix, and initial radium is set as d/10, step-length d/100.
6. according to claim 3 a kind of based on the camera one-dimension calibration method for improving PSO, it is characterised in that:Step S25
In, the radius of the higher-dimension ball is adjusted as unit of a step-length;It converts sphere center position and traverses entire square according to the method for browsing
The row element of battle array.
7. according to claim 1 a kind of based on the camera one-dimension calibration method for improving PSO, it is characterised in that:The step
Suddenly S3 is specially:The initial value of principal point coordinate is modified using existing priori physical message and according to multiple camera calibration
Parameter matrix mean square deviation, generate 8 dimension particle search range;Wherein, existing priori physical message includes:Phase owner
Immediate vicinity of the point in imaging sensor.
8. according to claim 1 a kind of based on the camera one-dimension calibration method for improving PSO, it is characterised in that:Step S4
Specifically include following steps:
Step S41:Initialize PSO;
Step S42:The adaptive value of each particle is calculated according to the following formula:
In formula, xpiWith ypiRespectively indicate the transverse and longitudinal coordinate of re-projection point, xiWith yiThe true transverse and longitudinal coordinate of subpoint is respectively indicated,
N indicates population number;
Step S43:If current adaptive value is better than the optimal adaptation value of particle, more new individual optimal adaptation valueIf fruit
Current Population adaptation value is better than the history optimal adaptation degree of population, then Population Regeneration optimal adaptation value
Step S44:The number of iterations adds one, and judges whether to reach the number of iterations or terminate threshold value, no if then terminating to optimize
Then, return step S43.
9. according to claim 8 a kind of based on the camera one-dimension calibration method for improving PSO, it is characterised in that:The step
Suddenly S41 is specially:Specify population number N, the number of iterations G, inertia weight w, acceleration constant c1、c2With end threshold value.
10. according to claim 9 a kind of based on the camera one-dimension calibration method for improving PSO, it is characterised in that:Population
Number N=30, the number of iterations G=500, inertia weight w=0.8, acceleration constant c1=1.2, c2=2.3.
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 true CN108898636A (en) | 2018-11-27 |
CN108898636B 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) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110853104A (en) * | 2020-01-15 | 2020-02-28 | 广东博智林机器人有限公司 | Calibration plate, machine vision calibration device and method |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101839692A (en) * | 2010-05-27 | 2010-09-22 | 西安交通大学 | Method for measuring three-dimensional position and stance of object with single camera |
CN102376093A (en) * | 2011-10-13 | 2012-03-14 | 清华大学 | Calibration method of camera |
JP2013093787A (en) * | 2011-10-27 | 2013-05-16 | Secom Co Ltd | 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 |
-
2018
- 2018-06-08 CN CN201810583687.4A patent/CN108898636B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101839692A (en) * | 2010-05-27 | 2010-09-22 | 西安交通大学 | Method for measuring three-dimensional position and stance of object with single camera |
CN102376093A (en) * | 2011-10-13 | 2012-03-14 | 清华大学 | Calibration method of camera |
JP2013093787A (en) * | 2011-10-27 | 2013-05-16 | Secom Co Ltd | 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 |
Non-Patent Citations (3)
Title |
---|
LI DENG ET AL.: "A novel camera calibration technique based on differential evolution particle swarm optimization algorithm", 《NEUROCOMPUTING》 * |
WEI GUAN ET AL.: "Improved Particle Swarm Optimization Algorithm Based NonLinear Calibration of Camera", 《IEEE》 * |
王德超等: "基于改进粒子群优化算法的非线性摄像机标定", 《计算机工程与科学》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110853104A (en) * | 2020-01-15 | 2020-02-28 | 广东博智林机器人有限公司 | Calibration plate, machine vision calibration device and method |
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 |
Also Published As
Publication number | Publication date |
---|---|
CN108898636B (en) | 2022-03-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109241895B (en) | Dense crowd counting method and device | |
CN109829437B (en) | Image processing method, text recognition device and electronic system | |
CN110322453B (en) | 3D point cloud semantic segmentation method based on position attention and auxiliary network | |
CN110659565B (en) | 3D multi-person human body posture estimation method based on porous convolution | |
CN105654476B (en) | Binocular calibration method based on Chaos particle swarm optimization algorithm | |
CN101908231A (en) | Reconstruction method and system for processing three-dimensional point cloud containing main plane scene | |
CN110147710B (en) | Method and device for processing human face features and storage medium | |
CN108960404B (en) | Image-based crowd counting method and device | |
CN108182421A (en) | Methods of video segmentation and device | |
WO2020113412A1 (en) | Target detection method and system | |
CN105405104B (en) | A kind of method and device of face image correcting | |
CN102170526A (en) | Method for calculation of defocus fuzzy core and sharp processing of defocus fuzzy image of defocus fuzzy core | |
CN110910330A (en) | Image processing method, image processing apparatus, electronic device, and readable storage medium | |
CN104881657B (en) | Side face recognition methods, side face construction method and system | |
CN105447869B (en) | Camera self-calibration method and device based on particle swarm optimization algorithm | |
CN116309844A (en) | Three-dimensional measurement method based on single aviation picture of unmanned aerial vehicle | |
CN108898636A (en) | A kind of camera one-dimension calibration method based on improvement PSO | |
CN111178503A (en) | Mobile terminal-oriented decentralized target detection model training method and system | |
CN112907972B (en) | Road vehicle flow detection method and system based on unmanned aerial vehicle and computer readable storage medium | |
CN112288813B (en) | Pose estimation method based on multi-view vision measurement and laser point cloud map matching | |
US8750641B2 (en) | Apparatus and method for correcting distortion of image | |
CN106327441A (en) | Image radial distortion automatic correction method and system | |
CN116167921B (en) | Method and system for splicing panoramic images of flight space capsule | |
CN110120009B (en) | Background blurring implementation method based on salient object detection and depth estimation algorithm | |
CN114611635B (en) | Object identification method and device, storage medium and electronic device |
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 |