CN105678740B - A kind of camera geometric calibration processing method and processing device - Google Patents
A kind of camera geometric calibration processing method and processing device Download PDFInfo
- Publication number
- CN105678740B CN105678740B CN201511015634.5A CN201511015634A CN105678740B CN 105678740 B CN105678740 B CN 105678740B CN 201511015634 A CN201511015634 A CN 201511015634A CN 105678740 B CN105678740 B CN 105678740B
- Authority
- CN
- China
- Prior art keywords
- value
- parameter
- camera
- cos distance
- iterative
- 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
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Studio Devices (AREA)
Abstract
The embodiment of the present invention provides a kind of camera geometric calibration processing method and processing device, the described method includes: determining camera parameter initial value, and the first COS distance of the point pair gone out based on the calculation of initial value is obtained, the point is to the coordinate to correspond to two pixels of space coordinate same position in two adjacent pictures；Judge whether first COS distance is consistent with preset value, if it is not, then obtaining the first iterative parameter, and adjusts the initial value using first iterative parameter, obtain the first optimal value；Obtain the second COS distance based on the calculated point pair of first optimal value；Judge whether second COS distance is consistent with the preset value, if it is, first optimal value is determined as camera parameter calibration value.Such scheme can substantially reduce calculation amount involved in treatment process, help to realize real-time online processing.
Description
Technical field
The present invention relates to field of image processing, in particular to a kind of camera geometric calibration processing method and processing device.
Background technique
Pan-shot is typically referred to carry out horizontal 360-degree and the shooting of vertical 180 degree centered on some point, will be captured
Plurality of pictures be spliced into shooting and the picture joining method of Zhang Quanjing's picture.In general, pan-shot at least may include
Two kinds of forms of panoramic picture and panoramic video.
In general, can be related to mapping and splice when being spliced into Zhang Quanjing's picture using multiple captured original images
Two parts.Wherein, mapping can be understood as projecting the pixel on original image on the corresponding position of panoramic pictures, splice
It can be understood as carrying out fusion transition to the overlapping region of adjacent two original images.
In order to determine the three-dimensional geometry position of space object surface point with it in the phase in original image between corresponding points
Mutual relation, can by way of camera geometric calibration, obtain camera parameter, so as to it is subsequent can use the camera parameter into
The projection of row pixel.In general, camera parameter may include the outer ginseng (e.g., roll, yaw, pitch, Tx, Ty, Tz) and camera of camera
Internal reference (such as u, v, w, α, beta, gamma).Wherein, (Tx, Ty, Tz) indicates that translation vector, (roll, yaw, pitch) indicate spin moment
Battle array, respectively representing around camera coordinates system z-axis rotation angle is γ, and rotating angle around y-axis is β, and rotating around x axis angle is α；(u,
V, w) it indicates to be biased to distortion, (α, beta, gamma) indicates panorama picture of fisheye lens model parameter.
Currently, it is iterated calculating using Levenberg-Marquardt algorithm (may be simply referred to as L-M algorithm) mostly, it is real
Existing camera geometric calibration.In which, each iterative process requires to seek second order local derviation to each parameter to be estimated, and obtains black plug
(Hessian) matrix and Jacobi (Jacobi) matrix.When carrying out pan-shot using N number of camera, calibration process is related to 12*
N number of parameter to be estimated, calculation amount is huge, needs to expend a large amount of calculating time, is usually used in servers off-line (offline) calibration,
It also cannot achieve real-time online processing at present.
In addition, when carrying out geometric calibration based on L-M algorithm, if occurring two determinants of a matrix in iterative process is zero,
I.e. transformation matrix is unusual, then it is believed that having obtained the local optimum of camera parameter.In general, local optimum can camera subject parameter
The influence of initial value, different initial values may result in that occur transformation matrices in different iterative process unusual, to obtain
Different local optimums, therefore, there is also higher requirements for selection of the which to initial value.
Summary of the invention
The embodiment of the present invention provides a kind of camera geometric calibration processing method and processing device, can substantially reduce involved by treatment process
And calculation amount, help to realize real-time online processing.
A kind of camera geometric calibration processing method, which comprises
It determines camera parameter initial value, and obtains the first COS distance of the point pair gone out based on the calculation of initial value, institute
It states a little to the coordinate to correspond to two pixels of space coordinate same position in two adjacent pictures；
Judge whether first COS distance is consistent with preset value, if it is not, then obtaining the first iterative parameter, and utilizes
First iterative parameter adjusts the initial value, obtains the first optimal value；
Obtain the second COS distance based on the calculated point pair of first optimal value；
Judge whether second COS distance is consistent with the preset value, if it is, first optimal value is true
It is set to camera parameter calibration value.
Preferably, if second COS distance is not inconsistent with the preset value, the method also includes:
Secondary iteration parameter is obtained, and adjusts first optimal value using the secondary iteration parameter, it is excellent to obtain second
Change value；
Obtain the third COS distance based on the calculated point pair of second optimal value；
Judge whether the third COS distance is consistent with the preset value, if it is, second optimal value is true
It is set to camera parameter calibration value.
Preferably, the determining camera parameter initial value, comprising:
The priori value of the camera parameter is determined as the initial value；Alternatively,
Increase random perturbation on the basis of the priori value of the camera, obtains the initial value.
Preferably, the mode of iterative parameter is obtained are as follows:
Machine learning is carried out to default sample, obtains the iterative parameter.
Preferably, machine learning is carried out to default sample, obtains the mode of first iterative parameter are as follows:
Wherein, (M0, N0) indicate the first iterative parameter, CjIndicate camera identification number,Indicate the camera of j-th of camera
Parameter calibration value,Indicate the camera parameter initial value of j-th of camera,It indicates to be based on (M0, N0) calculated point pair
First COS distance.
Preferably, described to adjust the initial value using first iterative parameter, obtain the mode of the first optimal value
Are as follows:
A kind of camera geometric calibration processing unit, described device include:
COS distance computing unit for determining camera parameter initial value, and obtains to go out based on the calculation of initial value
First COS distance of point pair, the point is to two pixels to correspond to space coordinate same position in two adjacent pictures
Coordinate；
Optimal value adjustment unit, for judging whether first COS distance is consistent with preset value, if it is not, then obtaining
First iterative parameter, and the initial value is adjusted using first iterative parameter, obtain the first optimal value；
The COS distance computing unit is also used to obtain based on the calculated point pair of first optimal value
Two COS distances；
Calibration value determination unit, for judging whether second COS distance is consistent with the preset value, if it is,
First optimal value is determined as camera parameter calibration value.
Preferably, described device further include:
The optimal value adjustment unit is also used to when second COS distance and the preset value are not inconsistent, and obtains the
Two iterative parameters, and first optimal value is adjusted using the secondary iteration parameter, obtain the second optimal value；
The COS distance computing unit is also used to obtain based on the calculated point pair of second optimal value
Three COS distances；
The calibration value determination unit, is also used to judge whether the third COS distance is consistent with the preset value, such as
Fruit is that second optimal value is then determined as camera parameter calibration value.
Preferably, described device further include:
Iterative parameter obtaining unit, for carrying out machine learning to default sample, acquisition first iterative parameter:
Wherein, (M0, N0) indicate the first iterative parameter, CjIndicate camera identification number,Indicate the camera of j-th of camera
Parameter calibration value,Indicate the camera parameter initial value of j-th of camera,It indicates to be based on (M0, N0) calculated point pair
First COS distance.
Preferably, the optimal value adjustment unit is specifically used for basisIt calculates and obtains first
Optimal value
Compared with prior art, the present invention program is fitted the mapping relations in picture splicing using Taylor polynomial,
Convert the optimization problem of iconic model parameter to the Nonlinear Convex double optimization problem of error between multinomial and mapping function.
Specifically, pixel projection can be carried out based on camera parameter current, the COS distance between institute's subpoint pair is obtained, further according to remaining
Chordal distance judges whether to need to carry out camera parameter optimization, if it is not required, then can demarcate parameter current as camera parameter
Value；If it is required, then obtaining iterative parameter using machine learning optimizes parameter current, and changed using the parameter after optimization
In generation, calculates, until obtaining camera parameter calibration value.Such scheme can substantially reduce calculation amount involved in treatment process,
Help to realize real-time online processing.
Detailed description of the invention
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for
For those of ordinary skill in the art, without any creative labor, it can also be obtained according to these attached drawings
His attached drawing.
Fig. 1 is the flow chart of camera geometric calibration processing method of the present invention；
Fig. 2 is the effect display diagram for the picture splicing realized based on the present invention program；
Fig. 3 is the effect display diagram for the picture splicing realized based on prior art；
Fig. 4 is the structural schematic diagram of camera geometric calibration processing unit of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Before introducing concrete scheme of the present invention, simple introduction first is done to mentality of designing of the present invention.
When carrying out pan-shot, there can be the region that partly overlaps in original image captured by any two adjacent cameras, it is right
This region that partly overlaps carries out fusion transition, can get panoramic pictures.Corresponding to this, we are it is to be understood that for overlay region
For domain, there are following scenes: there are a pixel a in two adjacent original image A and B, picture A, there are one in picture B
A pixel b, and a and b both correspond to the same position in space coordinate.Corresponding to this, pixel a and b can be known as
One point pair.
In splicing, it can be based on Current camera parameter, be respectively mapped to the pixel a and b on original image
In Equirectangular (equidistant spherical surface) projected image, the point pair that a and b corresponds on Equirectangular image is obtained
A ' and b ', in general, can reach best splicing effect if a ' and b ' can be overlapped.As an example, three-dimensional space can be calculated
Between COS distance between middle vector a ' and b ', the gap between a ' and b ' can be reflected by the COS distance, and then
Learn whether Current camera parameter is suitable for panoramic pictures splicing.
That is, can judge that whether Current camera parameter is suitable, if appropriate, then can be used it by COS distance
Carry out panoramic pictures splicing；If improper, camera parameter optimization can be carried out, is spelled to determine can be used in panoramic pictures
The camera parameter calibration value connect.In the present invention program, judge whether camera parameter suitable, it can be understood as COS distance whether with
Preset value is consistent, i.e., whether COS distance falls into the range of preset value permission.For example, preset value can be 0.1, lead to
Often, preset value is smaller, that is to say, that COS distance more levels off to 0, the panoramic pictures that the calibration value obtained based on optimization is spliced into
Effect is better, the embodiment of the present invention to the specific value of preset value without limitation, can be by practical application depending on.
To sum up, we can convert panorama camera geometric calibration process to one gradient direction step delta x of searching and solve
Following optimization problem:
Wherein,
aiAnd biIndicate i-th point pair of adjacent original image overlapping region；
H (*) indicates a function formed by camera parameter, to be realized based on camera parameter from the point on original image
The coordinate transform of point on to equidistant spherical surface image, e.g., aiBe converted to ai’, biBe converted to bi’；According to practical application, h (*) is usual
It can be presented as diversified forms, the present invention is to this and is not specifically limited, as long as coordinate conversion can be carried out based on camera parameter；
C (*, *) is indicated to solve the COS distance of two points on equidistant spherical surface image, such as be put to ai’And bi’COS distance.
The present invention program is illustrated below.
With reference to Fig. 1, the flow chart of camera geometric calibration processing method of the embodiment of the present invention is shown, may include following step
It is rapid:
S101, determines camera parameter initial value, and obtain the first cosine of the point pair gone out based on the calculation of initial value away from
From the point is to the coordinate to correspond to two pixels of space coordinate same position in two adjacent pictures.
The present invention is by way of judging whether COS distance a little pair falls into preset value allowed band, to determine whether needing
Camera parameter optimization is carried out, therefore, the initial value of a camera parameter can be first chosen, and a mapping is carried out based on this initial value
When penetrating, and then seeking out corresponding to initial value, the first COS distance of the point pair.
For example, the initial value in the present invention program can be set in conjunction with practical operation experience；Alternatively, considering camera
The priori value of parameter has usually been preferably to be worth, therefore the initial value that can be also arranged according to priori value in the present invention program e.g. can
The priori value of camera parameter is determined as initial value；Alternatively, random perturbation can be increased on the basis of the priori value of camera,
Obtain initial value.For example, random perturbation value can be arranged according to practical operation experience, alternatively, can also be by random perturbation value
± (priori value/100) are set as, the embodiment of the present invention can be not specifically limited this.
For example, CjIndicate the identification number of j-th of camera when pan-shot,Indicate the camera ginseng of j-th of camera
Number initial value, aJ, iAnd bJ, iIt indicates i-th point pair on original image captured by j-th of camera, can incite somebody to actionSubstitute into formula f
(x), the first COS distance d of the point pair gone out based on calculation of initial value is obtained1。
S102, judges whether first COS distance is consistent with preset value, if it is not, then the first iterative parameter is obtained,
And the initial value is adjusted using first iterative parameter, obtain the first optimal value.
Obtain d1Afterwards, the judgement for once whether optimizing camera parameter can be executed, that is, judge d1Whether it is consistent with preset value.
Still by taking preset value is 0.1 as an example, the first COS distance is consistent with preset value can be understood as d1≤ 0.1, otherwise it is assumed that more than first
Chordal distance is not consistent with preset value.
If through judgement think that camera parameter is needed to optimize, can from predefine iterative parameter set in,
The first iterative parameter is obtained, and using the initial value in the first iterative parameter adjustment S101, obtains the first optimization of camera parameter
Value.
For example, following formula optimization camera parameters can be passed through:Wherein,
(Mk-1, Nk-1) indicate k-th of iterative parameter,Indicate that j-th of camera needs optimised camera parameter,Indicate j-th of phase
Camera parameter after machine optimization.
In this step, the first optimal value(M0, N0) indicate the first iterative parameter,It indicates
The camera parameter initial value of j-th of camera,It indicates to be based on (M0, N0) calculated point pair the first COS distance d1。
In the present invention program, iterative parameter set can be obtained by way of carrying out machine learning to default sample, led to
Often, every progress an iteration just needs one group of iterative parameter.The mode for obtaining iterative parameter for the present invention can be found in hereafter institute
It introduces, wouldn't be described in detail herein.
S103 obtains the second COS distance based on the calculated point pair of first optimal value.
S104, judges whether second COS distance is consistent with the preset value, if it is, described first is optimized
Value is determined as camera parameter calibration value.
It is similar with S101, obtain the first optimal valueAfterwards, it can incite somebody to actionIt substitutes into formula f (x), obtains and correspond to first
When optimal value, the second COS distance d of the point pair2, and then recycle d2The judgement for once whether optimizing camera parameter executed.
Specifically, if judging result indicates d2It is consistent with preset value, then can terminates iterative process, by the first optimal valueReally
It is set to the parameter calibration value of j-th of camera, when carrying out pixel mapping based on the calibration value, the splicing effect of panoramic pictures can be made
Reach best.
In summary, the present invention is using the mapping relations in Taylor polynomial fitting picture splicing, by iconic model
The optimization problem of parameter is converted into the Nonlinear Convex double optimization problem of error between multinomial and mapping function.Based on the present invention
The camera geometric calibration that scheme is realized is not required to treat such as existing scheme and estimates parameter and seek second order local derviation, can substantially reduce treatment process
Related calculation amount helps to realize real-time online processing.In addition, the present invention is based on the fitting optimizations of the algorithm of machine learning to ask
Gradient descent direction in topic also to be simplified in process flow, helps speed up convergence rate, promotes the receipts of optimization problem
Hold back efficiency.In addition, being not easy to fall into local optimum based on the camera parameter calibration value that the present invention program determines, make present invention side
The optimization precision of case is higher.
Optionally, if the judging result of S104 indicates d2It is not inconsistent with preset value, then illustrates also to need by further changing
Camera parameter calibration value is obtained for process.Specifically, available secondary iteration parameter, and utilize the secondary iteration parameter
First optimal value is adjusted, the second optimal value is obtained；Obtain the based on the calculated point pair of second optimal value
Three COS distances；Judge whether the third COS distance is consistent with the preset value, if it is, by second optimal value
It is determined as camera parameter calibration value.In this example, the mode of the second optimal value is obtained, calculate third cosine using the second optimal value
The mode of distance judges mode that whether third COS distance be consistent with preset value, carries out subsequent processing according to judging result
Mode etc. can refer to and introduce at S102~S104 above, and details are not described herein again.
The mode for obtaining iterative parameter set to the present invention below does simple introduction.
Mode one can obtain iterative parameter set by way of machine learning.
Specifically, by the thought of the supervised learning in machine learning, learn each iteration institute out using training sample
Iterative parameter set { the M needed0, M1..., Mk-1, MkAnd { N0, N1..., Nk-1, Nk}。
Parameter involved in training sample can embody as follows:
A series of identity coding { the C of cameras for pan-shotj}；
Camera parameter calibration valueThe camera parameter calibration value for indicating j-th of camera is a 12*n dimensional vector；
Series of points is to { aJ, i, bJ, i, aJ, iAnd bJ, iIndicate i-th point on original image captured by j-th of camera
It is right；
Camera parameter initial valueIndicate the camera parameter initial value of j-th of camera.
In conjunction with above-mentioned sample parameter, the first iterative parameter (M can be obtained by solving following linear optimization problem0,
N0):
Obtain (M0, N0) after, it can foundationThe first optimal value is calculatedMeanwhile also
It can incite somebody to actionIt substitutes into formula f (x), calculates and obtainAnd then the secondary iteration parameter in the present invention program is obtained according to following formula
(M1, N1):
It is constantly calculated referring to aforesaid way, can determine iterative parameter set needed for the present invention program.For example, it passes through
It can be obtained when above-mentioned learning process discovery k=5 preferably as a result, after then learning, can get following iterative parameter set
{M0, M1, M2, M3, M4, M5And { N0, N1, N2, N3, N4, N5}.The present invention program does not do the value of k, value of iterative parameter etc.
It is specific to limit, can be by practical application depending on.
Mode two can obtain iterative parameter set in such a way that machine learning and parameter verification combine.
Specifically, it can be learnt according to process shown in mode one, and the iterative parameter collection that study obtains is collectively referred to as surveying
Try iterative parameter set.Then test sample is utilized, if the series of points of certain test camera is to { ai, bi, using present invention side
Case carries out camera parameter optimization using above-mentioned test iterative parameter set, and in k iteration since camera parameter initial value
Afterwards, point is obtained to { ai, biProjection { a on equidistant spherical surface imagei’, bi’, and then obtain the error parameter of the test camera.
For example, error parameter can be mean pixel error and maximum pixel error.
In general, if error parameter is consistent with preset threshold, then it is assumed that test iterative parameter set can obtained by the study stage
With determining it as the iterative parameter set in the present invention program；If error parameter is not inconsistent with preset threshold, then it is assumed that study
Test iterative parameter set obtained by stage is unavailable, corresponds to this, then can adjust the camera parameter initial value in study stage
And machine learning is carried out based on camera parameter initial value adjusted, until parameter verification stage obtained error parameter and default
Until threshold value is consistent.As an example, the camera parameter initial value in regularized learning algorithm stageCan be, in the base of initial value
Random perturbation is added on plinth.
In order to preferably verify beneficial effect brought by the present invention program, Experimental comparison results shown in following table are also provided.
It should be noted that the involved test environment of the verifying is intel i5 processor (3.3GHz), Win7 Pro system.
| Mean pixel error | Maximum pixel error | The number of iterations | Operation time (s) | |
| The present invention program | 1.85 | 5.20 | 5 | 3 |
| The prior art | 3.25 | 8.75 | 20 | 16 |
1 900 ten thousand pixel of table (2048 × 1536 × 3), three panorama camera calibration results
In addition to above-mentioned Experimental comparison results, from picture splicing effect, the present invention program is also better than the prior art,
For details, reference can be made to be based on the prior art shown in the effect display diagram for the picture splicing realized shown in Fig. 2 based on the present invention program, Fig. 3
The effect display diagram for the picture splicing that scheme is realized, institute's collar region especially in figure.
With method as described above correspondingly, the embodiment of the present invention also provides a kind of camera geometric calibration processing unit, ginseng
See Fig. 4, described device can include:
COS distance computing unit 201 for determining camera parameter initial value, and is obtained and is gone out based on the calculation of initial value
Point pair the first COS distance, the point in two adjacent pictures correspond to space coordinate same position two pixels
The coordinate of point；
Optimal value adjustment unit 202, for judging whether first COS distance is consistent with preset value, if it is not, then
The first iterative parameter is obtained, and adjusts the initial value using first iterative parameter, obtains the first optimal value；
The COS distance computing unit 201 is also used to obtain based on the calculated point pair of first optimal value
The second COS distance；
Calibration value determination unit 203, for judging whether second COS distance is consistent with the preset value, if
It is that first optimal value is then determined as camera parameter calibration value.
Optionally, described device further include:
The optimal value adjustment unit is also used to when second COS distance and the preset value are not inconsistent, and obtains the
Two iterative parameters, and first optimal value is adjusted using the secondary iteration parameter, obtain the second optimal value；
The COS distance computing unit is also used to obtain based on the calculated point pair of second optimal value
Three COS distances；
The calibration value determination unit, is also used to judge whether the third COS distance is consistent with the preset value, such as
Fruit is that second optimal value is then determined as camera parameter calibration value.
Optionally, described device further include:
Iterative parameter obtaining unit, for carrying out machine learning to default sample, acquisition first iterative parameter:
Wherein, (M0, N0) indicate the first iterative parameter, CjIndicate camera identification number,Indicate the camera of j-th of camera
Parameter calibration value,Indicate the camera parameter initial value of j-th of camera,It indicates to be based on (M0, N0) calculated point pair
First COS distance.
Optionally, the optimal value adjustment unit is specifically used for basisIt calculates and obtains first
Optimal value
It should be noted that all the embodiments in this specification are described in a progressive manner, each embodiment weight
Point explanation is the difference from other embodiments, and the same or similar parts between the embodiments can be referred to each other.
For device class embodiment, since it is basically similar to the method embodiment, so being described relatively simple, related place ginseng
See the part explanation of embodiment of the method.
Finally, it is to be noted that, herein, the terms "include", "comprise" or its any other variant are intended to
Cover non-exclusive inclusion, so that the process, method, article or equipment for including a series of elements not only includes those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or setting
Standby intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
There is also other identical elements in the process, method, article or apparatus that includes the element.
Scheme provided by the present invention is described in detail above, specific case used herein is to of the invention
Principle and embodiment is expounded, method and its core of the invention that the above embodiments are only used to help understand
Thought；At the same time, for those skilled in the art, according to the thought of the present invention, in specific embodiment and application range
Upper there will be changes, in conclusion the contents of this specification are not to be construed as limiting the invention.
Claims (5)
1. a kind of camera geometric calibration processing method, which is characterized in that the described method includes:
It determines camera parameter initial value, and obtains the first COS distance of the point pair gone out based on the calculation of initial value, the point
To in two adjacent pictures in original image captured by any two adjacent cameras correspond to space coordinate same position two
The coordinate of a pixel；
Judge whether first COS distance is consistent with preset value, if it is not, then from the iterative parameter set of going out is predefined
The first iterative parameter of middle acquisition, and the initial value is adjusted using first iterative parameter, obtain the first optimal value；
Obtain the second COS distance based on the calculated point pair of first optimal value；
Judge whether second COS distance is consistent with the preset value, if it is, first optimal value is determined as
Camera parameter calibration value；
Wherein, the mode of iterative parameter set is obtained are as follows:
By the supervised learning in machine learning, iterative parameter set needed for learning each iteration out using default sample.
2. the method according to claim 1, wherein if second COS distance and the preset value not
Symbol, the method also includes:
Secondary iteration parameter is obtained from the iterative parameter set, and excellent using secondary iteration parameter adjustment described first
Change value obtains the second optimal value；
Obtain the third COS distance based on the calculated point pair of second optimal value；
Judge whether the third COS distance is consistent with the preset value, if it is, second optimal value is determined as
Camera parameter calibration value.
3. the method according to claim 1, wherein the determining camera parameter initial value, comprising:
The priori value of the camera parameter is determined as the initial value；Alternatively,
Increase random perturbation on the basis of the priori value of the camera, obtains the initial value.
4. a kind of camera geometric calibration processing unit, which is characterized in that described device includes:
COS distance computing unit for determining camera parameter initial value, and obtains the point pair gone out based on the calculation of initial value
The first COS distance, the point in two adjacent pictures in original image captured by any two adjacent cameras to correspond to
The coordinate of two pixels of space coordinate same position；
Optimal value adjustment unit, for judging whether first COS distance is consistent with preset value, if it is not, then from advance really
The first iterative parameter is obtained in the iterative parameter set made, and adjusts the initial value using first iterative parameter, is obtained
Obtain the first optimal value, wherein obtain the mode of iterative parameter set are as follows: by the supervised learning in machine learning, using default
Sample is come iterative parameter set needed for learning each iteration out；
The COS distance computing unit is also used to obtain more than second based on the calculated point pair of first optimal value
Chordal distance；
Calibration value determination unit, for judging whether second COS distance is consistent with the preset value, if it is, by institute
It states the first optimal value and is determined as camera parameter calibration value.
5. device according to claim 4, which is characterized in that described device further include:
The optimal value adjustment unit is also used to when second COS distance and the preset value are not inconsistent, from the iteration
Secondary iteration parameter is obtained in parameter sets, and adjusts first optimal value using the secondary iteration parameter, obtains second
Optimal value；
The COS distance computing unit is also used to obtain more than the third based on the calculated point pair of second optimal value
Chordal distance；
The calibration value determination unit, is also used to judge whether the third COS distance is consistent with the preset value, if so,
Second optimal value is then determined as camera parameter calibration value.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201511015634.5A CN105678740B (en) | 2015-12-30 | 2015-12-30 | A kind of camera geometric calibration processing method and processing device |
| PCT/CN2016/078908 WO2017113535A1 (en) | 2015-12-30 | 2016-04-08 | Method and apparatus for geometric calibration of camera |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201511015634.5A CN105678740B (en) | 2015-12-30 | 2015-12-30 | A kind of camera geometric calibration processing method and processing device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105678740A CN105678740A (en) | 2016-06-15 |
| CN105678740B true CN105678740B (en) | 2019-06-07 |
Family
ID=56298011
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201511015634.5A Active CN105678740B (en) | 2015-12-30 | 2015-12-30 | A kind of camera geometric calibration processing method and processing device |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN105678740B (en) |
| WO (1) | WO2017113535A1 (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106875451B (en) * | 2017-02-27 | 2020-09-08 | 安徽华米智能科技有限公司 | Camera calibration method and device and electronic equipment |
| CN108320026B (en) * | 2017-05-16 | 2022-02-11 | 腾讯科技（深圳）有限公司 | Machine learning model training method and device |
| CN109166152A (en) * | 2018-07-27 | 2019-01-08 | 深圳六滴科技有限公司 | Bearing calibration, system, computer equipment and the storage medium of panorama camera calibration |
| CN110378963A (en) * | 2018-12-04 | 2019-10-25 | 北京京东振世信息技术有限公司 | Camera parameter scaling method and device |
| CN109598715B (en) * | 2018-12-05 | 2023-03-24 | 山西镭谱光电科技有限公司 | Material granularity online detection method based on machine vision |
| CN109785395A (en) * | 2019-01-16 | 2019-05-21 | 深圳市卓简科技有限公司 | A kind of panorama model parameter optimization method and system |
| CN110910458B (en) * | 2019-12-19 | 2022-12-02 | 长光禹辰信息技术与装备(青岛)有限公司 | Gray board calibration method, device, equipment and medium |
| CN113393529B (en) * | 2020-03-12 | 2024-05-10 | 浙江宇视科技有限公司 | Method, device, equipment and medium for calibrating camera |
| CN112102417B (en) * | 2020-09-15 | 2024-04-19 | 阿波罗智联(北京)科技有限公司 | Method and device for determining world coordinates |
| CN113405455B (en) * | 2021-05-17 | 2023-03-07 | 中创智维科技有限公司 | Method for automatically calibrating pixel ratio by calculating geometric parameters of contact network |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101794442A (en) * | 2010-01-25 | 2010-08-04 | 哈尔滨工业大学 | Calibration method for extracting illumination-insensitive information from visible images |
| CN101957990A (en) * | 2010-08-13 | 2011-01-26 | 武汉大学 | Camera calibration method, image processing equipment and motor vehicle |
| CN102881016A (en) * | 2012-09-19 | 2013-01-16 | 中科院微电子研究所昆山分所 | Vehicle 360-degree surrounding reconstruction method based on internet of vehicles |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102208108B (en) * | 2011-04-01 | 2013-01-09 | 哈尔滨工业大学 | Large-field high-precision quick field overall calibration method for camera |
| US10033989B2 (en) * | 2013-07-05 | 2018-07-24 | Mediatek Inc. | Synchronization controller for multi-sensor camera device and related synchronization method |
| CN103745458B (en) * | 2013-12-26 | 2015-07-29 | 华中科技大学 | A kind of space target rotating axle based on binocular light flow of robust and mass center estimation method |
| CN104751486B (en) * | 2015-03-20 | 2017-07-11 | 安徽大学 | A kind of moving target relay tracking algorithm of many ptz cameras |
-
2015
- 2015-12-30 CN CN201511015634.5A patent/CN105678740B/en active Active
-
2016
- 2016-04-08 WO PCT/CN2016/078908 patent/WO2017113535A1/en active Application Filing
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101794442A (en) * | 2010-01-25 | 2010-08-04 | 哈尔滨工业大学 | Calibration method for extracting illumination-insensitive information from visible images |
| CN101957990A (en) * | 2010-08-13 | 2011-01-26 | 武汉大学 | Camera calibration method, image processing equipment and motor vehicle |
| CN102881016A (en) * | 2012-09-19 | 2013-01-16 | 中科院微电子研究所昆山分所 | Vehicle 360-degree surrounding reconstruction method based on internet of vehicles |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2017113535A1 (en) | 2017-07-06 |
| CN105678740A (en) | 2016-06-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105678740B (en) | A kind of camera geometric calibration processing method and processing device | |
| EP3648055B1 (en) | Camera parameter calibration method, device, apparatus, and system | |
| CN105453131B (en) | The adjust automatically of image | |
| CN108257183B (en) | Camera lens optical axis calibration method and device | |
| CN105118055B (en) | Camera position amendment scaling method and system | |
| US9734427B2 (en) | Surveillance systems and image processing methods thereof | |
| CN105701837A (en) | Geometric calibration processing method and apparatus for camera | |
| CN103854291B (en) | Camera marking method in four-degree-of-freedom binocular vision system | |
| CN105654476B (en) | Binocular calibration method based on Chaos particle swarm optimization algorithm | |
| CN104778656B (en) | Fisheye image correcting method based on spherical perspective projection | |
| CN111553939B (en) | Image registration algorithm of multi-view camera | |
| CN111667398B (en) | Image processing method, apparatus and computer readable storage medium | |
| WO2014110954A1 (en) | Method, device and computer-readable storage medium for panoramic image completion | |
| CN101697105A (en) | Camera type touch detection positioning method and camera type touch detection system | |
| CN108171735B (en) | Billion pixel video alignment method and system based on deep learning | |
| CN105825470A (en) | Fisheye image correction method base on point cloud image | |
| CN106570907B (en) | Camera calibration method and device | |
| CN111340737B (en) | Image correction method, device and electronic system | |
| CN111445537A (en) | Calibration method and system of camera | |
| CN108492263A (en) | A kind of camera lens Lens Distortion Correction method | |
| CN106485781A (en) | Three-dimensional scene construction method based on live video stream and its system | |
| CN115563732A (en) | Spraying track simulation optimization method and device based on virtual reality | |
| CN104036496B (en) | Self-calibration method for radial distortion of fish-eye lens camera | |
| CN103793891A (en) | Low-complexity panorama image joint method | |
| CN112950528A (en) | Certificate posture determining method, model training method, device, server and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |