CN109781014A - The technology and methods of polyphaser collaboration on-line measurement strip target length under machine vision mode - Google Patents
The technology and methods of polyphaser collaboration on-line measurement strip target length under machine vision mode Download PDFInfo
- Publication number
- CN109781014A CN109781014A CN201910180429.6A CN201910180429A CN109781014A CN 109781014 A CN109781014 A CN 109781014A CN 201910180429 A CN201910180429 A CN 201910180429A CN 109781014 A CN109781014 A CN 109781014A
- Authority
- CN
- China
- Prior art keywords
- image
- target
- camera
- polyphaser
- strip
- 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
Abstract
The invention discloses the technology and methods of polyphaser collaboration on-line measurement strip target length under the machine vision mode of machine vision mode technical field, object to be measured image is shot using picture pick-up device such as industrial camera, length information of the target in picture pick-up device coordinate system is obtained with the ways and means of image procossing again, this method requires object to be measured panorama must be in captured image, when object to be measured overlength, its panorama can not be taken with a camera, polyphaser collaboration is needed to take pictures, collaboration processing is possible to out result;The present invention passes through when polyphaser shoots a super large target jointly, the photo of each camera can not simple concatenation restore the target, each image processing result does not have additive property, linear measure longimetry to strip overlength target, the result nonadditivity under polyphaser collaboration is overcome, while greatly reducing system processes data amount.
Description
Technical field
The present invention relates to machine vision mode technical fields, and in particular to polyphaser collaboration is online under machine vision mode surveys
Measure the technology and methods of strip target length.
Background technique
The production of current complex industrial product, which generally requires multiple link processings, to be completed, in not having also for each link processing
There is the processed goods as finished product to be known as middleware.It may need to understand middleware in time in these links of product working process
Dimensional information, it is desirable that in the case where not influencing the condition of production to the dimensional information for the middleware being on each processing links production line into
Row on-line measurement.
The advantage that machine vision mode has some other measurement methods not have in terms of on-line measurement, if do not contacted, can
To facilitate far from measurement target, later maintenance and equipment replacement, be particularly suitable for measurement high temperature, be not suitable for close or contact,
The target of movement.The image that industrial camera shoots middleware online is usually used in machine vision mode on-line measurement, then right
The image is handled in time obtains its dimensional information, and dimensional information is finally fed back to production line.If middleware is strip and ruler
It spends very big, its panorama can not be photographed using an industrial camera, a kind of method is to need to take pictures using more industrial cameras collaborations,
Then the seamless spliced panoramic picture for forming middleware is carried out to these cameras image of taking pictures, then it is obtained to these image procossings
Length information.The present invention is to carry out on-line measurement to strip target length information using machine vision mode under polyphaser collaboration
Method.
Long scheme problem to be solved is surveyed in polyphaser collaboration following problems: first is that the problem of machine vision aspect, i.e.,
Object to be measured segmentation and length information obtain in each camera image;Second is that each camera result how to reprocess acquisition meet it is actual
Total length information;Third is that the increase of camera quantity leads to the sharp increase of length measurement system processing data volume, system burden can also increase sharply,
The timeliness of this what state guarantee on-line measurement.The content of present invention is laid particular emphasis on to the polyphaser association in the measurement of strip target length
The solution of the latter two problems faced in.Based on this, the present invention devises the online survey of polyphaser collaboration under machine vision mode
The technology and methods of strip target length are measured, to solve the above problems.
Summary of the invention
The purpose of the present invention is to provide the skills of polyphaser collaboration on-line measurement strip target length under machine vision mode
How art and method reprocess acquisition to solve each camera result mentioned above in the background art and meet actual total length letter
The problem of increase of breath and camera quantity leads to the sharp increase of length measurement system processing data volume, and system burden can also increase sharply.
To achieve the above object, the invention provides the following technical scheme: the online survey of polyphaser collaboration under machine vision mode
Measure the technology and methods of strip target length, it is assumed that piece image is f (x, y), and the operation for extracting attribute to image procossing is set as
H is transformed to T by measurement coordinate system to world coordinate system, and the actual result to the processing of diagram picture is R, and measurement process can be with
It is described as
R=T (H (f (x, y))) (1)
Core is the definition and realization of H in this process;In the measuring system of view-based access control model, H (f (x, y)) result
The distance between two measuring points are namely based on, the two measuring point coordinates are based on image coordinate system, since distance is to seat
Mark transformation has invariance, therefore long system is surveyed for one camera, and (1) formula becomes
R=H (f (x, y)) (2)
For the length measurement system of polyphaser collaboration, the H of each camera is identical, measurement result R, testing image f (x, y) and seat
It is different to mark transformation matrix, i-th of camera is represented by
Ri=Ti(H(fi(x,y))) (3)
For entire polyphaser cooperative system, comprehensive each camera result is that the last measurement result of whole system also needs one
A operation;It is assumed that the operation that comprehensive each camera result is the last measurement result of whole system is Tc, result is R after synthesis, then more
Camera cooperates with lower measurement process can be described as
R=Tc(Ti(H(fi(x,y)))) (4)
Assuming that entire measuring system is made of n camera, the operation that attribute is extracted to image procossing is H, and system has n survey
Coordinate system is measured, each measurement coordinate system corresponds to a transformation and is set as Ti(i=0,1,2,3,4 ..., n-1), the figure that camera is taken pictures
As fi(x, y) (i=0,1,2,3,4 ..., n-1), obtaining actual result to image procossing is Ri(i=0,1,2,3,4 ..., n-
1), final measurement R then has
Ri=Ti(H(fi(x,y))) (5)
R=Tc(R0,R1, R=..., Rn-1) (6)
By above-mentioned two formula combine as
R=Tc(T0(H(f0(x,y))),…,Ti(H(fi(x,y))),…,Tn-1(H(fn-1(x,y)))) (7)
According to above problem model, it needs to be determined that there is T in the measuring system of polyphaser collaborationcAnd Ti(i=0,1,2,
3,4 ..., n-1) two kinds of transformation and image procossing H;
There is the T under the ideally polyphaser collaboration of additive property for each camera resultcTransformation and TiConvert two changes
It is as follows to change matrix setting
S in formula (8)xi、SyiThe respectively transformation coefficient of coordinate x, y, X0i、Y0iFor image fiThe measurement coordinate system of (x, y)
Starting point, the coordinate of corresponding world coordinate system, is generated, T by HcIt is the unit matrix of n dimension, it is assumed that H (fi(x, y) result is
(Xi,Yi), then formula (7) becomes
Measuring length is
As previously described, above scheme establishment is conditional, i.e., each camera horizontal direction is consistent and picture is not overlapped,
This is difficult to accomplish when realizing in practice, furthermore the program measures every time must handle n width image, obtains in n picture
After length, then their results are added up, how many camera, which participates in collaboration, will handle how many width images, and that does not lighten the burden is excellent
Gesture;For reality, how T is definedcTransformation, TiTransformation and H needs are studied, when designing, establishing conversion scheme, side
As assessment foundation in terms of the practical operability and the algorithm speed of service two of case.
Preferably, it redefines and G is operated to image, which is that the product of N and newly-increased operation D is operated to image,
I.e.
G=ND or DN (13)
It is that H and coordinate system transformation T was operated to image originally to image operation NiProduct, i.e., not only obtained to after image procossing
Strip target endpoint coordinate, but also will transformation to from the coordinate to world coordinate system, i.e.,
N=TiH (14)
Newly-increased operation D is defined as follows
ci=D (fi(x, y)) (15)
ciIt is a constant, constant ciValue is as follows
(7) formula above in this way can be further written as
R=Tc(T0((c0)H(f0(x,y))),…,Ti((ci)H(fi(x,y))),…,Tn-1((cn-1)H(fn-1(x,
y)))) (16)
Due to ciTo TiFor constant, it is possible to move on to TiBefore transformation, i.e.,
R=Tc(c0T0(H(f0(x,y))),…,ciT=(H (fi(x,y))),…,cn-1Tn-1(H(fn-1(x,y))))
(17)
To any one strip target, only there are two camera c in industrial camera queueiIt is not 0, other is 0, if strip
Target both ends are located at i-th and j-th of camera, then above formula becomes
R=Tc(-(ci)Ti(H(fi(x,y))),(cj)Tj(H(fj(x,y)))) (18)
Become in new coordinate transform (10), (11) formula
Rx=(SxjXj-X0j)-(SxiXi-X0i) (19)
Ry=(SyjYj-Y0j)-(SyiYi-Y0i) (20)
It redefines after operating G to image, all increases each image a D operation, it appears that it is multiple to increase system processing
Miscellaneous degree, actually significantly reduces data processing amount, although to be D operation, but T to each camera imageiOperation and H operation
Do not need to carry out each image, every time measurement only to two images carry out can, as long as each image carry out D operation
The two edges of check image can determine, does not need to check whole image, greatly accelerates algorithm speed.
Preferably, what is redefined is the composition operation of multiple operations to image operation G, including the D operation to image, H
Operation and Ti transformation;The realization for operating G is mainly the realization of D operation, H operation;For piece image, D operation is main to be answered
Strip target be left end or right end in the picture, i.e., generation constant ci, H operation is that the endpoint location of strip target to be obtained is sat
Whether in the picture the two requirements are summed up the both ends for seeking to determine strip target by mark, if will further really
Surely be left end or right end in the images, specific coordinate position, this coordinate position must also be the value of world coordinate system;
It can be divided into three classes the case where polyphaser cooperates with situation, and strip target occurs in the image of polyphaser, it is a kind of
Strip target left end in image, the second class be strip target without endpoint in the picture, the only segment section of strip target scheming
As in, third class is strip target right end in image;For any one strip target, the first kind under polyphaser cooperative mode
Respectively there was only piece image with third class image, the second class image there may be several, it is also possible to not have;
1) each camera investigates the realization of region setting and H operation
To industry spot on-line measurement, the position that object to be measured occurs in each camera image is not arbitrary, but is gone out
A present fixed area, target region is different between each camera, but the target region for some certain camera
Be it is certain, to make full use of this feature when image operation G is realized in research, propose region of interest ROI or investigate area
Domain concept, it is the region that target to be measured occurs in the picture that investigating region, which is exactly the fixed area in image, and setting is investigated
The advantage in region is that data processing amount is greatly decreased, and avoids the processing of complex background, so that algorithm design is simplified;
The principle of region setting is investigated first is that the treating capacity of data is reduced to the greatest extent, second is that making Processing Algorithm simple possible;It examines
Examine region setting shape, size etc. and application in relation to, it is related with target shape to be split, this is grown for the survey of strip target
Business, investigation region are arranged in the region that strip target is passed through and strip, through image the right and left as far as possible, are arranged in this way
After investigating region afterwards, investigate region and be divided into two parts, strip target internal point and outer portion, the two should attribute it is different, hand over
Be exactly strip target left end at boundary, as long as region is investigated in each search sweep in this way, investigate range be greatly reduced, complex background pair
The influence of algorithm can not almost consider;H operation can only investigate region in carry out, can using threshold method or other
Partition means determine the boundary of target and background, and at this moment coordinate is still based on the coordinate system of image;
2) D operation realization and the complete algorithm of image procossing
H operation is the extreme coordinates for returning to strip target, and D operation is constant c to be obtainedi, i.e., whether there is or not strips in the image
Target endpoint and differentiation are left end point or right endpoint, as long as D operation investigates the left and right in region close to the two of image border
Whether the two column pixels at end are that target pixel points are obtained with answer, and speed can be quickly.
H operation and D operate the complete algorithm being combined, and constant c is arranged according to two column situation of ROI left and right edgesiStep
It is as follows
1. when two column are target object pixel, ciConstant=0;
2. when two column are not target object pixel, ciConstant=0;
3. right column are not target pixel points, c when left column is target pixel pointsiConstant=1;
4. when the right side is classified as target pixel points, left column is not target pixel points, ciConstant=- 1;
3) image calibration and coordinate transform
Image is after D is operated, if ciConstant ≠ 0, then H operation can provide the strip target side based on camera coordinates system
Point coordinate, the extreme coordinates have to pass through coordinate transform, and the coordinate for converting it into world coordinate system could become measured value, sit
Mark transformation is it needs to be determined that Sxi、Syi、X0i、Y0i, in polyphaser collaboration, have multiple camera coordinates systems, how to determine these ginsengs
Number is also to need to study;
Under polyphaser collaboration, in order to realize conversion of the camera coordinates system to world coordinate system, need to each camera point
It is not demarcated, to obtain the S of each cameraxi、Syi、X0i、Y0i, the physics of specific calibration and object to be measured shape and measurement
Measure related, calibration is exactly to demarcate it point by point in world coordinates along target length direction for strip target surveys this long application
The length of system.
Preferably, the two edges of the two edges of the D operation inspection image are and strip target vertical direction.
Preferably, described image calibration and calibration process in coordinate transform can first determine the calibration point in the picture, so
It the point is corresponded to background again afterwards really puts and determine its world coordinate system coordinate value, be also possible to for real background
In specified point look for its corresponding diagram picture point in the picture, the distribution of calibration request point must cover entire mesh in the longitudinal direction
Mark range.
Compared with prior art, the beneficial effects of the present invention are: the present invention by shooting a super large in polyphaser jointly
When target, the photo of each camera can not simple concatenation restore the target, each image processing result does not have additive property, to strip
The linear measure longimetry of overlength target overcomes the result nonadditivity under polyphaser collaboration, while greatly reducing system processing number
According to amount.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, will be described below to embodiment required
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for ability
For the those of ordinary skill of domain, without creative efforts, it can also be obtained according to these attached drawings other attached
Figure.
Fig. 1 is polyphaser synergetic structure example schematic of the present invention.
Fig. 2 is the more picture spliced panoramic schematic diagrames of polyphaser of the present invention.
Fig. 3 is the practical polyphaser image mosaic example schematic of the present invention.
Fig. 4 is the three classes situation schematic diagram that strip target endpoint of the present invention is distributed in the picture.
Fig. 5 is that the present invention investigates region setting example schematic.
Fig. 6 lists intention for D of the present invention operation is investigated two.
Fig. 7 is the complete algorithm flow schematic diagram of the present invention.
Fig. 8 is present invention calibration schematic diagram.
Fig. 9 is that present system implements architecture diagram schematic diagram.
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 all other
Embodiment shall fall within the protection scope of the present invention.
Please refer to Fig. 1-9, the present invention provides a kind of technical solution: polyphaser cooperates with on-line measurement item under machine vision mode
The technology and methods of shape target length, it is assumed that piece image is f (x, y), and the operation for extracting attribute to image procossing is set as H, by
It measures coordinate system and is transformed to T to world coordinate system, the actual result to the processing of diagram picture is R, and measurement process can be retouched
State for
R=T (H (f (x, y))) (1)
Core is the definition and realization of H in this process;In the measuring system of view-based access control model, H (f (x, y)) result
The distance between two measuring points are namely based on, the two measuring point coordinates are based on image coordinate system, since distance is to seat
Mark transformation has invariance, therefore long system is surveyed for one camera, and (1) formula becomes
R=H (f (x, y)) (2)
For the length measurement system of polyphaser collaboration, the H of each camera is identical, measurement result R, testing image f (x, y) and seat
It is different to mark transformation matrix, i-th of camera is represented by
Ri=Ti(H(fi(x,y))) (3)
For entire polyphaser cooperative system, comprehensive each camera result is that the last measurement result of whole system also needs one
A operation;It is assumed that the operation that comprehensive each camera result is the last measurement result of whole system is Tc, result is R after synthesis, then more
Camera cooperates with lower measurement process can be described as
R=Tc(Ti(H(fi(x,y)))) (4)
Assuming that entire measuring system is made of n camera, the operation that attribute is extracted to image procossing is H, and system has n survey
Coordinate system is measured, each measurement coordinate system corresponds to a transformation and is set as Ti(i=0,1,2,3,4 ..., n-1), the figure that camera is taken pictures
As fi(x, y) (i=0,1,2,3,4 ..., n-1), obtaining actual result to image procossing is Ri(i=0,1,2,3,4 ..., n-
1), final measurement R then has
Ri=Ti(H(fi(x,y))) (5)
R=Tc(R0,R1, R=..., Rn-1) (6)
By above-mentioned two formula combine as
R=Tc(T0(H(f0(x,y))),…,Ti(H(fi(x,y))),…,Tn-1(H(fn-1(x,y)))) (7)
According to above problem model, it needs to be determined that there is T in the measuring system of polyphaser collaborationcAnd Ti(i=0,1,2,
3,4 ..., n-1) two kinds of transformation and image procossing H;
There is the T under the ideally polyphaser collaboration of additive property for each camera resultcTransformation and TiConvert two changes
It is as follows to change matrix setting
S in formula (8)xi、SyiThe respectively transformation coefficient of coordinate x, y, X0i、Y0iFor image fiThe measurement coordinate system of (x, y)
Starting point, the coordinate of corresponding world coordinate system, is generated, T by HcIt is the unit matrix of n dimension, it is assumed that H (fi(x, y) result is
(Xi,Yi), then formula (7) becomes
Measuring length is
As previously described, above scheme establishment is conditional, i.e., each camera horizontal direction is consistent and picture is not overlapped,
This is difficult to accomplish when realizing in practice, furthermore the program measures every time must handle n width image, obtains in n picture
After length, then their results are added up, how many camera, which participates in collaboration, will handle how many width images, and that does not lighten the burden is excellent
Gesture;For reality, how T is definedcTransformation, TiTransformation and H needs are studied, when designing, establishing conversion scheme, side
As assessment foundation in terms of the practical operability and the algorithm speed of service two of case.
Wherein, it redefines and G is operated to image, which is to operate the product of N and newly-increased operation D to image, i.e.,
G=ND or DN (13)
It is that H and coordinate system transformation T was operated to image originally to image operation NiProduct, i.e., not only obtained to after image procossing
Strip target endpoint coordinate, but also will transformation to from the coordinate to world coordinate system, i.e.,
N=TiH (14)
Newly-increased operation D is defined as follows
ci=D (fi(x, y)) (15)
ciIt is a constant, constant ciValue is as follows
(7) formula above in this way can be further written as
R=Tc(T0((c0)H(f0(x,y))),…,Ti((ci)H(fi(x,y))),…,Tn-1((cn-1)H(fn-1(x,
y)))) (16)
Due to ciTo TiFor constant, it is possible to move on to TiBefore transformation, i.e.,
R=Tc(c0T0(H(f0(x,y))),…,ciT=(H (fi(x,y))),…,cn-1Tn-1(H(fn-1(x,y))))
(17)
To any one strip target, only there are two camera c in industrial camera queueiIt is not 0, other is 0, if strip
Target both ends are located at i-th and j-th of camera, then above formula becomes
R=Tc(-(ci)Ti(H(fi(x,y))),(cj)Tj(H(fj(x,y)))) (18)
Become in new coordinate transform (10), (11) formula
Rx=(SxjXj-X0j)-(SxiXi-X0i) (19)
Ry=(SyjYj-Y0j)-(SyiYi-Y0i) (20)
It redefines after operating G to image, all increases each image a D operation, it appears that it is multiple to increase system processing
Miscellaneous degree, actually significantly reduces data processing amount, although to be D operation, but T to each camera imageiOperation and H operation
Do not need to carry out each image, every time measurement only to two images carry out can, as long as each image carry out D operation
The two edges of check image can determine, does not need to check whole image, greatly accelerates algorithm speed.
What is redefined is the composition operation of multiple operations to image operation G, including D operation, H operation and the Ti to image
Transformation;The realization for operating G is mainly the realization of D operation, H operation;For piece image, D operation is main to answer strip target
Be left end or right end in the picture, i.e., generation constant ci, H operation is the endpoint location coordinate of strip target to be obtained, by this
Two require whether to be in the picture summed up the both ends for seeking to determine strip target, if to further determine that being left
In the images, specific coordinate position, this coordinate position must also be the value of world coordinate system for end or right end;
It can be divided into three classes the case where polyphaser cooperates with situation, and strip target occurs in the image of polyphaser, it is a kind of
Strip target left end in image, the second class be strip target without endpoint in the picture, the only segment section of strip target scheming
As in, third class is strip target right end in image;For any one strip target, the first kind under polyphaser cooperative mode
Respectively there was only piece image with third class image, the second class image there may be several, it is also possible to not have;
1) each camera investigates the realization of region setting and H operation
To industry spot on-line measurement, the position that object to be measured occurs in each camera image is not arbitrary, but is gone out
A present fixed area, target region is different between each camera, but the target region for some certain camera
Be it is certain, to make full use of this feature when image operation G is realized in research, propose region of interest ROI or investigate area
Domain concept, it is the region that target to be measured occurs in the picture that investigating region, which is exactly the fixed area in image, and setting is investigated
The advantage in region is that data processing amount is greatly decreased, and avoids the processing of complex background, so that algorithm design is simplified;
The principle of region setting is investigated first is that the treating capacity of data is reduced to the greatest extent, second is that making Processing Algorithm simple possible;It examines
Examine region setting shape, size etc. and application in relation to, it is related with target shape to be split, this is grown for the survey of strip target
Business, investigation region are arranged in the region that strip target is passed through and strip, through image the right and left as far as possible, are arranged in this way
After investigating region afterwards, investigate region and be divided into two parts, strip target internal point and outer portion, the two should attribute it is different, hand over
Be exactly strip target left end at boundary, as long as region is investigated in each search sweep in this way, investigate range be greatly reduced, complex background pair
The influence of algorithm can not almost consider;H operation can only investigate region in carry out, can using threshold method or other
Partition means determine the boundary of target and background, and at this moment coordinate is still based on the coordinate system of image;
2) D operation realization and the complete algorithm of image procossing
H operation is the extreme coordinates for returning to strip target, and D operation is constant c to be obtainedi, i.e., whether there is or not strips in the image
Target endpoint and differentiation are left end point or right endpoint, as long as D operation investigates the left and right in region close to the two of image border
Whether the two column pixels at end are that target pixel points are obtained with answer, and speed can be quickly.
H operation and D operate the complete algorithm being combined, and constant c is arranged according to two column situation of ROI left and right edgesiStep
It is as follows
1. when two column are target object pixel, ciConstant=0;
2. when two column are not target object pixel, ciConstant=0;
3. right column are not target pixel points, c when left column is target pixel pointsiConstant=1;
4. when the right side is classified as target pixel points, left column is not target pixel points, ciConstant=- 1;
3) image calibration and coordinate transform
Image is after D is operated, if ciConstant ≠ 0, then H operation can provide the strip target side based on camera coordinates system
Point coordinate, the extreme coordinates have to pass through coordinate transform, and the coordinate for converting it into world coordinate system could become measured value, sit
Mark transformation is it needs to be determined that Sxi、Syi、X0i、Y0i, in polyphaser collaboration, have multiple camera coordinates systems, how to determine these ginsengs
Number is also to need to study;
Under polyphaser collaboration, in order to realize conversion of the camera coordinates system to world coordinate system, need to each camera point
It is not demarcated, to obtain the S of each cameraxi、Syi、X0i、Y0i, the physics of specific calibration and object to be measured shape and measurement
Measure related, calibration is exactly to demarcate it point by point in world coordinates along target length direction for strip target surveys this long application
The length of system.
The two edges of the two edges of D operation inspection image are and strip target vertical direction.
Calibration process can first determine the calibration point in the picture in image calibration and coordinate transform, then again the point pair
It should really be put to background and be also possible to exist for the specified point in real background to determine its world coordinate system coordinate value
Its corresponding diagram picture point is looked in image, the distribution of calibration request point must cover entire target zone in the longitudinal direction.
One concrete application of the present embodiment are as follows: object to be measured image is shot using picture pick-up device such as industrial camera, then is used
The ways and means of image procossing obtain length information of the target in picture pick-up device coordinate system.This method requires mesh to be measured
Marking panorama must be in captured image.When object to be measured overlength, its panorama can not be taken with a camera, need
Polyphaser collaboration is taken pictures, collaboration processing is possible to out result.
Start first to investigate two column of ROI or so, constant c is arranged according to two column situation of ROIi, judge ciWhether constant is equal to 0, when
ciWhen constant is equal to 0, directly terminate;Work as ciIt when constant is not equal to 0, investigates inside ROI, obtains target side coordinate, is i.e. H is operated, so
After terminate.
Firstly the need of more industrial cameras, the resolution ratio coincidence measurement required precision of camera, camera must be GiGE interface
's;Secondly a computer is needed to run system of the invention;Finally need the equipment of some networkings.Each camera is passed through into net
Line is connected on the network card interface of computer.Guarantee that each camera and computer constitute an independent network;It again will be based on this
The software installation of invention is after computer, and it is long can to run survey for system after appropriately configured.It is shot jointly in polyphaser
When one super large target, the photo of each camera can not simple concatenation restore the target, each image processing result does not have and can add
Property.Linear measure longimetry to strip overlength target, the result that the present invention proposes that a kind of new method overcomes under polyphaser collaboration can not
Additivity, while greatly reducing system processes data amount.
In the description of this specification, the description of reference term " one embodiment ", " example ", " specific example " etc. means
Particular features, structures, materials, or characteristics described in conjunction with this embodiment or example are contained at least one implementation of the invention
In example or example.In the present specification, schematic expression of the above terms may not refer to the same embodiment or example.
Moreover, particular features, structures, materials, or characteristics described can be in any one or more of the embodiments or examples to close
Suitable mode combines.
Present invention disclosed above preferred embodiment is only intended to help to illustrate the present invention.There is no detailed for preferred embodiment
All details are described, are not limited the invention to the specific embodiments described.Obviously, according to the content of this specification,
It can make many modifications and variations.These embodiments are chosen and specifically described to this specification, is in order to better explain the present invention
Principle and practical application, so that skilled artisan be enable to better understand and utilize the present invention.The present invention is only
It is limited by claims and its full scope and equivalent.
Claims (5)
1. the technology and methods of polyphaser collaboration on-line measurement strip target length under machine vision mode, it is characterised in that: false
If piece image is f (x, y), the operation for extracting attribute to image procossing is set as H, by the change of measurement coordinate system to world coordinate system
It is changed to T, the actual result to the processing of diagram picture is R, and measurement process can be described as
R=T (H (f (x, y))) (1)
Core is the definition and realization of H in this process;In the measuring system of view-based access control model, H (f (x, y)) result is exactly
Based on the distance between two measuring points, the two measuring point coordinates are based on image coordinate system, since distance becomes coordinate
It changes with invariance, therefore surveys long system for one camera, (1) formula becomes
R=H (f (x, y)) (2)
For the length measurement system of polyphaser collaboration, the H of each camera is identical, measurement result R, testing image f (x, y) and coordinate change
It is different to change matrix, i-th of camera is represented by
Ri=Ti(H(fi(x,y))) (3)
For entire polyphaser cooperative system, comprehensive each camera result is that the last measurement result of whole system also needs a behaviour
Make;It is assumed that the operation that comprehensive each camera result is the last measurement result of whole system is Tc, result is R after synthesis, then polyphaser
Lower measurement process is cooperateed with to can be described as
R=Tc(Ti(H(fi(x,y)))) (4)
Assuming that entire measuring system is made of n camera, the operation for extracting attribute to image procossing is H, and system has n measurement to sit
Mark system, each measurement coordinate system correspond to a transformation and are set as Ti(i=0,1,2,3,4 ..., n-1), the image f that camera is taken picturesi
(x, y) (i=0,1,2,3,4 ..., n-1), obtaining actual result to image procossing is Ri(i=0,1,2,3,4 ..., n-1), most
Whole measurement result is R, then has Ri=Ti(H(fi(x,y))) (5)
R=Tc(R0,R1, R=..., Rn-1) (6)
By above-mentioned two formula combine as
R=Tc(T0(H(f0(x,y))),…,Ti(H(fi(x,y))),…,Tn-1(H(fn-1(x,y)))) (7)
According to above problem model, it needs to be determined that there is T in the measuring system of polyphaser collaborationcAnd Ti(i=0,1,2,3,
4 ..., n-1) two kinds of transformation and image procossing H;
There is the T under the ideally polyphaser collaboration of additive property for each camera resultcTransformation and TiConvert two transformation squares
Battle array setting is as follows
S in formula (8)xi、SyiThe respectively transformation coefficient of coordinate x, y, X0i、Y0iFor image fiThe measurement coordinate system of (x, y) originates
Point, the coordinate of corresponding world coordinate system, is generated, T by HcIt is the unit matrix of n dimension, it is assumed that H (fi(x, y) result is (Xi,
Yi), then formula (7) becomes
Measuring length is
As previously described, above scheme establishment is conditional, i.e., each camera horizontal direction is consistent and picture is not overlapped, this
It is difficult to accomplish when realizing in practice, furthermore the program measure every time must handle n width image, the length in n picture of acquisition
Afterwards, then their results it adds up, how many camera, which participates in collaboration, will handle how many width images, the advantage that do not lighten the burden;
For reality, how T is definedcTransformation, TiTransformation and H needs are studied, when designing, establishing conversion scheme, scheme
As assessment foundation in terms of practical operability and the algorithm speed of service two.
2. under machine vision mode according to claim 1 polyphaser collaboration on-line measurement strip target length technology and
Method, it is characterised in that: it redefines and G is operated to image, which is that the product of N and newly-increased operation D is operated to image,
I.e.
G=ND or DN (13)
It is that H and coordinate system transformation T was operated to image originally to image operation NiProduct, i.e., to not only obtaining strip after image procossing
Target endpoint coordinate, but also will transformation to from the coordinate to world coordinate system, i.e.,
N=TiH (14)
Newly-increased operation D is defined as follows
ci=D (fi(x, y)) (15)
ciIt is a constant, constant ciValue is as follows
(7) formula above in this way can be further written as
R=Tc(T0((c0)H(f0(x, y))) ..., Ti((ci)H(fi(x, y))) ..., Tn-1((cn-1)H(fn-1(x, y))))
(16)
Due to ciTo TiFor constant, it is possible to move on to TiBefore transformation, i.e.,
R=Tc(c0T0(H(f0(x, y))) ..., ciT=(H (fi(x, y))) ..., Cn-1Tn-1(H(fn-1(x, y))))
(17)
To any one strip target, only there are two camera c in industrial camera queueiIt is not 0, other is 0, if strip target
Both ends are located at i-th and j-th of camera, then above formula becomes
R=Tc(-(ci)Ti(H(fi(x, y))), (cj)Tj(H(fj(x, y)))) (18)
Become in new coordinate transform (10), (11) formula
Rx=(SxjXj-X0j)-(SxiXi-X0i) (19)
Ry=(SyjYj-Y0j)-(SyiYi-Y0i) (20)
It redefines after operating G to image, all increases each image a D operation, it appears that system processing complexity is increased,
But data processing amount is actually significantly reduced, although to be D operation, T to each camera imageiOperation and H operation are not required to
Each image is carried out, every time measurement only to two images carry out can, as long as each image carry out D operation check
The two edges of image can determine, does not need to check whole image, greatly accelerates algorithm speed.
3. under machine vision mode according to claim 1 polyphaser collaboration on-line measurement strip target length technology and
Method, it is characterised in that: what is redefined is the composition operation of multiple operations to image operation G, including the D operation to image, H
Operation and Ti transformation;The realization for operating G is mainly the realization of D operation, H operation;For piece image, D operation is main to be answered
Strip target be left end or right end in the picture, i.e., generation constant ci, H operation is that the endpoint location of strip target to be obtained is sat
Whether in the picture the two requirements are summed up the both ends for seeking to determine strip target by mark, if will further really
Surely be left end or right end in the images, specific coordinate position, this coordinate position must also be the value of world coordinate system;
It can be divided into three classes the case where polyphaser cooperates with situation, and strip target occurs in the image of polyphaser, one kind is item
Shape target left end in image, the second class be strip target without endpoint in the picture, only strip target segment section in the picture,
Third class is strip target right end in image;For any one strip target, the first kind and under polyphaser cooperative mode
Three classes image respectively only has piece image, and the second class image may have several, it is also possible to not have;
1) each camera investigates the realization of region setting and H operation
To industry spot on-line measurement, the position that object to be measured occurs in each camera image is not arbitrary, and is occurred from
One fixed area, target region is different between each camera, but target region is one for some certain camera
Fixed, this feature is made full use of when image operation G is realized in research, proposes that region of interest ROI or investigation region are general
It reads, it is the region that target to be measured occurs in the picture that investigating region, which is exactly the fixed area in image, and region is investigated in setting
Advantage be that data processing amount is greatly decreased, avoid the processing of complex background so that algorithm design simplify;
The principle of region setting is investigated first is that the treating capacity of data is reduced to the greatest extent, second is that making Processing Algorithm simple possible;Investigate area
Domain be arranged shape, size etc. and application in relation to, it is related with target shape to be split, this task is grown for the survey of strip target,
The region that strip target is passed through and strip is arranged in, through image the right and left in investigation region as far as possible, examines after being arranged in this way
After examining region, investigate region and be divided into two parts, strip target internal point and outer portion, the two should attribute it is different, intersection
Exactly strip target left end, as long as region is investigated in search sweep each in this way, range is greatly reduced, complex background is to algorithm for investigation
Influence can not almost consider;H operation only can investigate progress in region, can use threshold method or other segmentations
Means determine the boundary of target and background, and at this moment coordinate is still based on the coordinate system of image;
2) D operation realization and the complete algorithm of image procossing
H operation is the extreme coordinates for returning to strip target, and D operation is constant c to be obtainedi, i.e., whether there is or not strip target sides in the image
Put and distinguish left end point or right endpoint, if D operation investigate investigate region left and right close to image border both ends two
Whether column pixel is that target pixel points are obtained with answer, and speed can be quickly.
H operation and D operate the complete algorithm being combined, and constant c is arranged according to two column situation of ROI left and right edgesiSteps are as follows
1. when two column are target object pixel, ciConstant=0;
2. when two column are not target object pixel, ciConstant=0;
3. right column are not target pixel points, c when left column is target pixel pointsiConstant=1;
4. when the right side is classified as target pixel points, left column is not target pixel points, ciConstant=- 1;
3) image calibration and coordinate transform
Image is after D is operated, if ciConstant ≠ 0, then H operation can provide the strip target endpoint based on camera coordinates system and sit
Mark, the extreme coordinates have to pass through coordinate transform, and the coordinate for converting it into world coordinate system could become measured value, and coordinate becomes
It changes it needs to be determined that Sxi、Syi、X0i、Y0i, in polyphaser collaboration, have multiple camera coordinates systems, how to determine these parameters
Needs are studied;
Polyphaser collaboration under, in order to realize conversion of the camera coordinates system to world coordinate system, need to each camera respectively into
Rower is fixed, to obtain the S of each cameraxi、Syi、X0i、Y0i, specific calibration and the physical quantity of object to be measured shape and measurement have
It closes, calibration is exactly to demarcate it point by point in world coordinate system along target length direction for strip target surveys this long application
Length.
4. under machine vision mode according to claim 2 polyphaser collaboration on-line measurement strip target length technology and
Method, it is characterised in that: the two edges of the two edges of the D operation inspection image are and strip target vertical direction.
5. under machine vision mode according to claim 3 polyphaser collaboration on-line measurement strip target length technology and
Method, it is characterised in that: calibration process can first determine the calibration point in the picture in described image calibration and coordinate transform, so
It the point is corresponded to background again afterwards really puts and determine its world coordinate system coordinate value, be also possible to for real background
In specified point look for its corresponding diagram picture point in the picture, the distribution of calibration request point must cover entire mesh in the longitudinal direction
Mark range.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910180429.6A CN109781014B (en) | 2019-03-11 | 2019-03-11 | Technology and method for online measuring length of strip-shaped target through cooperation of multiple cameras in machine vision mode |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910180429.6A CN109781014B (en) | 2019-03-11 | 2019-03-11 | Technology and method for online measuring length of strip-shaped target through cooperation of multiple cameras in machine vision mode |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109781014A true CN109781014A (en) | 2019-05-21 |
CN109781014B CN109781014B (en) | 2020-10-16 |
Family
ID=66488816
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910180429.6A Active CN109781014B (en) | 2019-03-11 | 2019-03-11 | Technology and method for online measuring length of strip-shaped target through cooperation of multiple cameras in machine vision mode |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109781014B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110278380A (en) * | 2019-07-18 | 2019-09-24 | 成都甄识科技有限公司 | A kind of restructural super more mesh cameras and its multiplexing method |
CN112697065A (en) * | 2021-01-25 | 2021-04-23 | 东南大学 | Three-dimensional shape reconstruction method based on camera array |
CN112797900A (en) * | 2021-04-07 | 2021-05-14 | 中科慧远视觉技术(北京)有限公司 | Multi-camera plate size measuring method |
Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1286134A1 (en) * | 2001-08-18 | 2003-02-26 | Trumpf Werkzeugmaschinen GmbH + Co. KG | Apparatus and method for measuring the geometry of workpieces |
CN2773676Y (en) * | 2005-04-04 | 2006-04-19 | 中国印钞造币总公司 | Composite field registration detection housing with multiple cameras |
CN102042814A (en) * | 2010-06-24 | 2011-05-04 | 中国人民解放军国防科学技术大学 | Projection auxiliary photographing measurement method for three-dimensional topography of large storage yard |
US20120099119A1 (en) * | 1999-04-05 | 2012-04-26 | Faro Technologies Inc. | Laser-based coordinate measuring device and laser-based method for measuring coordinates |
CN102519400A (en) * | 2011-12-15 | 2012-06-27 | 东南大学 | Large slenderness ratio shaft part straightness error detection method based on machine vision |
CN103971353A (en) * | 2014-05-14 | 2014-08-06 | 大连理工大学 | Splicing method for measuring image data with large forgings assisted by lasers |
CN104021540A (en) * | 2013-02-28 | 2014-09-03 | 宝山钢铁股份有限公司 | Static state calibration device and method for machine visual surface detection equipment |
CN105823416A (en) * | 2016-03-04 | 2016-08-03 | 大族激光科技产业集团股份有限公司 | Method for measuring object through multiple cameras and device thereof |
CN205449822U (en) * | 2016-02-29 | 2016-08-10 | 鞍钢股份有限公司 | A polyphaser image mosaicking device for belted steel surface is detected |
CN105953730A (en) * | 2016-06-22 | 2016-09-21 | 首航节能光热技术股份有限公司 | Multi-camera solar heat collector steel structure support assembling quality detection system |
CN106886979A (en) * | 2017-03-30 | 2017-06-23 | 深圳市未来媒体技术研究院 | A kind of image splicing device and image split-joint method |
CN107091610A (en) * | 2017-04-19 | 2017-08-25 | 清华大学 | The Three-Dimensional Dynamic on-line measurement device and its measuring method of a kind of large scale structure |
CN109099883A (en) * | 2018-06-15 | 2018-12-28 | 哈尔滨工业大学 | The big visual field machine vision metrology of high-precision and caliberating device and method |
-
2019
- 2019-03-11 CN CN201910180429.6A patent/CN109781014B/en active Active
Patent Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120099119A1 (en) * | 1999-04-05 | 2012-04-26 | Faro Technologies Inc. | Laser-based coordinate measuring device and laser-based method for measuring coordinates |
EP1286134A1 (en) * | 2001-08-18 | 2003-02-26 | Trumpf Werkzeugmaschinen GmbH + Co. KG | Apparatus and method for measuring the geometry of workpieces |
CN2773676Y (en) * | 2005-04-04 | 2006-04-19 | 中国印钞造币总公司 | Composite field registration detection housing with multiple cameras |
CN102042814A (en) * | 2010-06-24 | 2011-05-04 | 中国人民解放军国防科学技术大学 | Projection auxiliary photographing measurement method for three-dimensional topography of large storage yard |
CN102519400A (en) * | 2011-12-15 | 2012-06-27 | 东南大学 | Large slenderness ratio shaft part straightness error detection method based on machine vision |
CN104021540A (en) * | 2013-02-28 | 2014-09-03 | 宝山钢铁股份有限公司 | Static state calibration device and method for machine visual surface detection equipment |
CN103971353A (en) * | 2014-05-14 | 2014-08-06 | 大连理工大学 | Splicing method for measuring image data with large forgings assisted by lasers |
CN205449822U (en) * | 2016-02-29 | 2016-08-10 | 鞍钢股份有限公司 | A polyphaser image mosaicking device for belted steel surface is detected |
CN105823416A (en) * | 2016-03-04 | 2016-08-03 | 大族激光科技产业集团股份有限公司 | Method for measuring object through multiple cameras and device thereof |
CN105953730A (en) * | 2016-06-22 | 2016-09-21 | 首航节能光热技术股份有限公司 | Multi-camera solar heat collector steel structure support assembling quality detection system |
CN106886979A (en) * | 2017-03-30 | 2017-06-23 | 深圳市未来媒体技术研究院 | A kind of image splicing device and image split-joint method |
CN107091610A (en) * | 2017-04-19 | 2017-08-25 | 清华大学 | The Three-Dimensional Dynamic on-line measurement device and its measuring method of a kind of large scale structure |
CN109099883A (en) * | 2018-06-15 | 2018-12-28 | 哈尔滨工业大学 | The big visual field machine vision metrology of high-precision and caliberating device and method |
Non-Patent Citations (1)
Title |
---|
万袁冬: "用于零件检测的多相机标定与图像拼接方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110278380A (en) * | 2019-07-18 | 2019-09-24 | 成都甄识科技有限公司 | A kind of restructural super more mesh cameras and its multiplexing method |
CN112697065A (en) * | 2021-01-25 | 2021-04-23 | 东南大学 | Three-dimensional shape reconstruction method based on camera array |
CN112697065B (en) * | 2021-01-25 | 2022-07-15 | 东南大学 | Three-dimensional shape reconstruction method based on camera array |
CN112797900A (en) * | 2021-04-07 | 2021-05-14 | 中科慧远视觉技术(北京)有限公司 | Multi-camera plate size measuring method |
CN112797900B (en) * | 2021-04-07 | 2021-07-06 | 中科慧远视觉技术(北京)有限公司 | Multi-camera plate size measuring method |
Also Published As
Publication number | Publication date |
---|---|
CN109781014B (en) | 2020-10-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
DE112012001984B4 (en) | Integrate video metadata into 3D models | |
DE112013003338B4 (en) | Size measuring device and size measuring method | |
Shiranita et al. | Determination of meat quality by texture analysis | |
EP2471040B1 (en) | Method and device for joining a plurality of individual digital images into a total image | |
DE102017001667A1 (en) | An image processing device for displaying an object that is recognized by an input image capture | |
CN109781014A (en) | The technology and methods of polyphaser collaboration on-line measurement strip target length under machine vision mode | |
DE112018001050T5 (en) | SYSTEM AND METHOD FOR VIRTUALLY ENHANCED VISUAL SIMULTANEOUS LOCALIZATION AND CARTOGRAPHY | |
DE112011100652T5 (en) | THREE-DIMENSIONAL MEASURING DEVICE, PROCESSING METHOD AND NON-VOLATILE COMPUTER-READABLE STORAGE MEDIUM | |
DE112018000332T5 (en) | SEAL VISUAL SLAM WITH PROBABILISTIC SURFEL MAP | |
DE112016004535T5 (en) | Universal Compliance Network | |
WO2019096952A1 (en) | A system and method for single image object density estimation | |
DE102013113490A1 (en) | Method and system for estimating a position of a camera | |
DE102018207414A1 (en) | Image processing system | |
Xian et al. | UprightNet: geometry-aware camera orientation estimation from single images | |
DE102018119682A1 (en) | Image processing device, image processing method and non-temporary computer readable memory | |
DE102013211240A1 (en) | Range measuring device and range measuring method | |
DE112018003571T5 (en) | Information processing device and information processing method | |
DE102018003475A1 (en) | Form-based graphic search | |
DE102010039652A1 (en) | Mosaic shot generation | |
CN108171735A (en) | 1,000,000,000 pixel video alignment schemes and system based on deep learning | |
Solorzano et al. | Whole slide image registration for the study of tumor heterogeneity | |
EP2528042A1 (en) | Method and device for the re-meshing of 3D polygon models | |
JP6900704B2 (en) | Cell observation system | |
Li et al. | Self-supervised coarse-to-fine monocular depth estimation using a lightweight attention module | |
DE102004026782A1 (en) | Method and apparatus for computer-aided motion estimation in at least two temporally successive digital images, computer-readable storage medium and computer program element |
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 |