CN109859272A - A kind of auto-focusing binocular camera scaling method and device - Google Patents
A kind of auto-focusing binocular camera scaling method and device Download PDFInfo
- Publication number
- CN109859272A CN109859272A CN201811550433.9A CN201811550433A CN109859272A CN 109859272 A CN109859272 A CN 109859272A CN 201811550433 A CN201811550433 A CN 201811550433A CN 109859272 A CN109859272 A CN 109859272A
- Authority
- CN
- China
- Prior art keywords
- camera
- binocular camera
- calibration
- calibrating
- calibrating template
- 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
Landscapes
- Length Measuring Devices By Optical Means (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses a kind of auto-focusing binocular camera scaling method and device and depth computing method, the scaling method includes the following steps: A1, obtains the calibrating template image at the respective distance shot at multiple calibration distances, each corresponding individual calibrating template image of calibration distance;Have size is known to demarcate pattern on the calibrating template;This is calculated apart from one group of corresponding binocular camera calibrating parameters apart from corresponding one individual calibrating template image in A2, analysis and each calibration of processing;Corresponding multiple calibration distances, are always obtained the calibrating parameters of multiple groups binocular camera.The present invention can be applied to the calibration of the camera (or camera) with automatic focusing function, have the advantages that method is simple, efficient.Additionally, it is provided the camera calibration at multiple distances, can better compensate for calibrating parameters difference because of caused by lens location variation, so that successive depths calculating is more acurrate.
Description
Technical field
The present invention relates to computer vision, optical measurement and camera manufacturing field more particularly to a kind of auto-focusing are double
Mesh camera calibration method and device.
Background technique
Auto-focusing refers to that one kind built in camera is automatically performed to shot subject focusing simultaneously by electronics and mechanical device
Image is set to reach clearly function.Due to having the characteristics that focusing is accurate, easy to operate, currently with the camera shooting of automatic focusing function
Head is all widely used in more and more industries, including the fields such as smart phone, unmanned plane, video monitoring.
On the other hand, camera application enters the 3D epoch based on binocular, more mesh cameras from 2D.Such as in mobile phone
Industry realizes optical zoom function by the combination of wide-angle camera and focal length camera, or passes through black and white and colored camera shooting
The combination of head promotes picture quality, or grabs the 3D point cloud of face using infrared structure light video camera head mould group, realizes people
Face identification.Unmanned plane industry realizes 3D environment sensing by binocular camera to realize unmanned plane avoidance awing.Garage
Industry completes ranging, 3D environment sensing by binocular camera to assist driving and supports automatic Pilot.
The calibration of camera is the key that realize the above binocular camera 3D application.In measurement and vision based on camera
In, three-dimensional position of the object point in space corresponds to it relationship between two-dimensional pixel point position in the picture, mathematically
It is described with geometric projection model.The parameter of the model is generally by calibration pattern (such as solid circular array known to size
Or black and white gridiron pattern) take pictures, image procossing and be calculated, this determines that the process of camera projection model parameter is referred to as
For calibration.These parameters include: intrinsic parameter, refer to principle point location c when camera imagingx、cyAnd focal length fx、fy, such parameter is only
It is related to camera itself;Outer parameter refers to that camera in the position in space, refers generally to camera in a certain reference coordinate system
Rotating vector R and translation vector T;Distortion parameter, during referring to that camera is taken pictures, object point practical respective pixel position in the picture
Deviation between the theoretical subpoint calculated based on imaging model, the deviation is generally by radial distortion parameter k1、k2、k3And it cuts
To distortion parameter p1、p2To describe.
When there are two limitations for the general calibration solution of preceding camera volume production production line.First, due to the calibration used
Algorithm requires to clap the images of multiple calibrating templates from different perspectives, perhaps rotated in current producing line 2D plane reference template or
Use the calibration volume template spliced by muti-piece plane reference template (having fixed angle from each other).Rotate the dress of calibrating template
Mechanism complexity is set, and rotates and has elongated time of measuring, yield in unit time is reduced, to increase cost.Muti-piece plane mark
The volume template of solid plate splicing perhaps needs accurately to fix or need by certain angle to make other than increased costs
With preceding high-acruracy survey volume template itself.
Camera with automatic focusing function, camera lens are generally fixed on an electronics and mechanical motion mechanism, example
Such as VCM motor (voice coil motor).Camera lens is pushed before and after optical axis in lens barrel by movement mechanism, changes position.Camera lens at
As the distance change (i.e. image distance is changed) of chip surface, the depth of field of camera focusing is also changed therewith, thus
The focus function for the different depth of field may be implemented.For the limitation of this function: the camera with auto-focusing, for each
A depth of field, camera lens have a corresponding position in lens barrel, its calibrating parameters value of different positions is also different.Theoretically, it takes the photograph
As head needs to re-scale each depth of field depth in each lens location.Current scaling scheme is not examined
Consider the variation of camera lens calibrating parameters caused by lens barrel change in location.
Summary of the invention
It is an object of the invention to propose a kind of auto-focusing binocular camera scaling method and device and depth calculation
Method, it is simple, efficient and accurate.
For this purpose, auto-focusing binocular camera scaling method of the invention includes the following steps: A1, obtains in multiple calibration
The calibrating template image at respective distance shot at distance, each corresponding individual calibrating template image of calibration distance;Institute
Stating on calibrating template has size is known to demarcate pattern;A2, analysis and processing it is each calibration apart from corresponding one individual
This is calculated apart from one group of corresponding binocular camera calibrating parameters in calibrating template image;A3, corresponding multiple calibration away from
From the calibrating parameters of multiple groups binocular camera are always obtained.
In some embodiments, the invention also includes following technical characteristics:
In step A1, the calibration pattern is solid circular array, which arranges a filled circles by C row × L and form,
C and L is natural number;Each filled circles size is identical, and radius is consistent, and both horizontally and vertically distance of center circle is all the same;Each away from
From the uniform cause of calibration pattern, be made of C × L filled circles;Each distance calibration pattern radius of circle and distance of center circle with
The distance of binocular camera to the calibrating template is directly proportional.
In step Al, two camera optical axises are each perpendicular to calibrating template when shooting the calibrating template image or so,
So that uncalibrated image focusing and there is no because of pattern deformation caused by shooting angle.
In step A2, analyze and handle it is each calibration apart from corresponding one individual calibrating template image specifically include as
Lower step: S2, in the calibrating template image that left and right camera is clapped passes through where image processing algorithm extracts the filled circles center of circle
Location of pixels;S3, it is based on the calibrating template image, calculates the focal length of each camera lens of binocular camera, obtains each camera internal reference
The initial value of several, outer parameter, distortion parameter;S4, it is based on Levenberg-Marquardt optimization algorithm, in binocular camera
Parameter, outer parameter, distortion parameter optimize, and obtain accurate binocular camera calibrating parameters;S5, calculate this apart from when this pair
The relative position parameter of mesh camera.
In step S2, filled circles region is detected based on gray value of image, seeks round homogeneous seat of the region mass center as the center of circle
It marks [x y 1].
In step S3, two intrinsic parameters, focal length f are solved by following equation groupxAnd fy, initial value:
h11h12·B11+(h31h12+h11h32)·B13+h21h22·B22+(h31h22+h21h32)·
B23+h31h32·B33=0
Wherein, hijIt is the homography matrix H-matrix element of the i-th row jth column (i, j=1,2,3), is known;
BijIt is determined by following formula:
In above formula, K is the intrinsic parameter of camera, is defined as
cxAnd cyIt is the principal point of image, presets principal point cxAnd cyInitial value be picture centre.
In step S4, when optimization, so that the quadratic sum of the re-projection error in all filled circles centers of circle is minimum on calibrating template,
That is:
Wherein M is the position in the filled circles center of circle on calibrating template, and m is that the centre point corresponds to actual pixels position in image,Attach most importance to projected position, i.e., the filled circles center of circle after central projection, because of figure
Location of pixels that image distortion and calculations of offset come out (AndDeviated comprising distortion), K is the intrinsic parameter of camera, rotating vector R
It is the outer parameter of camera, k=[k with translation vector T1k2k3] it is radial distortion parameter, p=[p1 p2] it is tangential distortion parameter.
In step S5, the relative position of two cameras itself is
Wherein, Rl、TlFor the spatial position of left camera, Rr、TrFor the spatial position of right camera;
If there is multiple cameras, then calculated according to this between any two.
The invention also includes a kind of auto-focusing binocular camera caliberating devices, which is characterized in that using as described above
Scaling method.
The invention also includes a kind of depth measurement methods, which is characterized in that is obtained pair using scaling method as described above
The calibrating parameters of the multiple groups binocular camera of multiple calibration distances are answered to select camera lens position when from current bat figure in depth test
It sets a nearest calibration and calculates depth apart from one group of corresponding binocular camera calibrating parameters.
The present invention can be applied to the calibration of the camera (or camera) with automatic focusing function, and due to only with a width
Calibrating template image, which carries out parameter calibration, has the advantages that method is simple, efficient compared to existing general scaling method.In addition,
The device provides the camera calibration at multiple distances, can better compensate for calibrating parameters and cause because of lens location variation
Difference, thus successive depths calculate it is more acurrate.
Detailed description of the invention
Fig. 1 is the structure chart of the binocular camera caliberating device of the embodiment of the present invention.
Fig. 2 is the calibrating template calibration pattern schematic diagram of the embodiment of the present invention.
Fig. 3 is the process step schematic diagram of caliberating device of embodiment of the present invention calibration binocular camera mould group.
Fig. 4 is the flow diagram of binocular camera mould group of embodiment of the present invention scaling method in each distance calibration.
Fig. 5 is depth information calculating schematic diagram for one point P of object scene of embodiment of the present invention surface.
Specific embodiment
Below by the calibration side for the binocular camera with automatic focusing function that specific embodiment illustrates the present invention
Method and device.It, need to (N be integer, can at N number of distance on the production line of binocular camera when being demarcated using this method
With the requirement applied according to subsequent binocular camera, the requirement of producing line yield in unit time and producing line equipment size size requirements etc. are suitable
Work as selection;This specification illustrates by taking four distances as an example, respectively 0.5,1.0,1.5,2.0 meters, similarly hereinafter) respectively put one piece of 2D plane
Calibrating template;Each calibration distance only claps the image of calibrating template at the distance, and the calibrating template does not need to rotate:
The present invention presets principal point cxAnd cyInitial value be picture centre, so that a figure can find out fxAnd fyThe two unknown numbers
Initial value, calibrating template do not need to rotate, due to during camera actual production, optical lens and imager chip coupling
The step of being combined dress requires the principal point of image from imager chip central point without departing from certain pixel coverage, and therefore, the present invention is this
The default using effect that will not influence it in actual production;Analyze and handle individual calibrating template figure clapped at the distance
This can be calculated apart from corresponding binocular camera calibrating parameters in picture;Corresponding multiple (four) calibration distances, are always obtained
The calibrating parameters of multiple groups (four groups) binocular camera;In depth test, lens location is nearest when from current bat figure one is selected
Calibrating parameters are organized to calculate depth.
Embodiment one:
Fig. 1 is the structure chart of caliberating device on binocular camera production line.Demarcating distance can be according to precision, production capacity, size
Equal requirements suitably select, and for this sentences four distances, respectively put the calibrating template of one piece of 2D plane in four distances: 0.5,1.0,
1.5,2.0 meters.
Calibrating template is as shown in Figure 2.This scaling scheme is using solid circular array as calibrating template pattern, the filled circles battle array
Column arrange a filled circles by C row × L and form, and C and L are natural number.Each filled circles size is identical, and radius is consistent, and horizontal and vertical
Histogram is all the same to distance of center circle.The uniform cause of calibration pattern of each distance, is made of C × L filled circles.The mark of each distance
Radius of circle and the distance of center circle for determining pattern are directly proportional at a distance from binocular camera to the calibrating template.
Fig. 1 middle circle represents a rotating mechanism, and binocular camera mould group is fixed on rotating mechanism, successively rotates to
The position of face calibrating template, i.e. binocular camera optical axis is perpendicular to calibrating template.After reaching the position, rotating mechanism stops
Movement, binocular camera static state clap figure, and calibration algorithm calculates camera calibration parameter.Rotational structure starts to act simultaneously, images
Head goes to next calibration position.
The step of caliberating device calibration binocular camera mould group disclosed by the invention, is as follows, as shown in Figure 3:
Step 1: mould group feeding to be measured;
Step 2: rotation calculates calibrating parameters at 1.5m to 1.5m calibrating template, calibrating procedure at 1.5m, is clapped;
Step 3: rotation calculates calibrating parameters at 0.5m to 0.5m calibrating template, calibrating procedure at 0.5m, is clapped;
Step 4: rotation calculates calibrating parameters at 2.0m to 2.0m calibrating template, calibrating procedure at 2.0m, is clapped;
Step 5: rotation calculates calibrating parameters at 1.0m to 1.0m calibrating template, calibrating procedure at 1.0m, is clapped;
Step 6: mould group blanking to be measured.
Wherein, in each distance calibration, the process of scaling method is as follows, as shown in Figure 4:
S1, obtain one this apart from when calibrating template image;
S2, in the calibrating template image that left and right camera is clapped, image processing algorithm extract the filled circles center of circle where
Location of pixels;
S3, it is based on the uncalibrated image, calculates the focal length of binocular camera camera lens, obtain camera internal reference number, outer parameter, abnormal
The initial value of variable element;
S4, it is based on Levenberg-Marquardt optimization algorithm, to binocular camera intrinsic parameter, outer parameter, distortion parameter
It optimizes, obtains accurate binocular camera calibrating parameters;
S5, calculate this apart from when the binocular camera relative position parameter.
Correspondingly, device used in above-mentioned calibrating method includes:
Figure unit is obtained, the calibrating template image obtained for obtaining a width by shooting calibrating template;
Extraction unit, for detecting filled circles in calibrating template image, to extract center pixel position;
Unit is demarcated, binocular camera intrinsic parameter, outer parameter (rotating vector and translation vector) and distortion are calculated or set
The initial value of parameter prepares the optimization of next step calibrating parameters;
Optimize unit and Levenberg-Marquardt is used based on the minimum of all centre point re-projection error quadratic sums
Algorithm optimizes the intrinsic parameter of binocular camera, outer parameter and distortion parameter, obtains accurate calibration result.
Position determination unit, the space bit based on previous step two cameras obtained by calibrating relative to same calibrating template
It sets, determines the relative position of binocular camera.
Wherein, in each distance calibration, detailed description are as follows for the process of scaling method:
1, the calibrating template image that a width is obtained by shooting calibrating template is obtained;(corresponding to S1)
Calibrating template is generally the repeat patterns for having constant spacing, such as black and white chessboard case marker solid plate, equidistant reality
Heart circular array calibrating template etc..
As shown in Fig. 2, this scaling scheme uses solid circular array as calibrating template pattern, the solid circular array is by 8 rows
× 11 arrange a filled circles composition.Each filled circles size is identical, and radius is consistent, and both horizontally and vertically distance of center circle is all the same.
The uniform cause of calibration pattern of each distance, is made of 8 × 11 filled circles.The radius of circle and circle of the calibration pattern of each distance
The heart is away from directly proportional at a distance from binocular camera to the calibrating template.
In this example, each calibration distance only needs to clap a calibrating template image, and left and right two when shooting
Camera optical axis is each perpendicular to calibrating template, so that uncalibrated image clear patterns and there is no because shooting is because of pattern caused by angle
Deformation.It is placed in LED panel lamp additionally, due to calibrating template, demarcates pattern and white background comparison is strong, demarcate pattern edge
Comparison is strong, is easy to extract.It is suitable that filled circles array calibrating template is used based on the above two o'clock favorable factor, in the present embodiment
's.
According to the number and distance of center circle for both horizontally and vertically going up filled circles in calibrating template, it may be determined that solid circular array exists
Distribution in calibrating template coordinate system obtains the homogeneous coordinates [X Y Z 1] in the center of circle.
2, solid loop truss is carried out to the calibrating template image, to extract centre point;(corresponding to S2)
In computer vision fields such as 3 D scene rebuildings, duplicate filled circles are frequently utilized that construct calibration pattern, are led to
Fixed radius of circle and distance of center circle are crossed to determine calibrating template size.The filled circles center of circle has easily detecting, position precision height, matching
Reliably, the advantages such as real-time processing.Current center of circle detection algorithm includes: that the center of circle based on Blob regional analysis is detected, based on side
The center of circle detection that edge extracts, center of circle detection based on Hough Hough transform etc..In this example, after the completion of uncalibrated image shooting,
Filled circles region is detected based on gray value of image, seeks round homogeneous coordinates [x y 1] of the region mass center as center pixel position.
The image processing step calculates simple, strong real-time.
3, calculate or set the first of binocular camera intrinsic parameter, outer parameter (rotating vector and translation vector) and distortion parameter
Value;(corresponding to S3)
In computer vision, on space object a little with its by imaging system on as plane projected position it is mutual
Relationship is generally described by the geometric projection model of camera (or camera) system.Common projection model is based in optics
The central projection of pinhole imaging system principle.In the model, projection centre, the i.e. optical center of camera lens are a little passed through on object, are thrown along straight line
Shadow is on imager chip.
The filled circles center of circle is [X Y Z 1] in the homogeneous coordinates of reference coordinate system, it is assumed that obtained by the point is taken pictures in camera
Pixel homogeneous coordinates be [x y 1].According to the projection model based on pinhole imaging system, the calibrating template filled circles center of circle [X Y 1]
It is projected on image by following relationship, obtains corresponding imaging pixel point [x y 1] (herein for the feelings of plane reference template
0) condition, Z coordinate are assumed to
Wherein σ is scale factor.Rotating vector R and translation vector T is the outer parameter of camera, and description camera is being demarcated
The spatial position of template coordinate system.K is the intrinsic parameter of camera, is defined as
Wherein fxAnd fyFor focal length both horizontally and vertically, cxAnd cyIt is the principal point of image.
The image of the calibrating template shot based on every, can calculate its corresponding homography matrix
Wherein hjIt is the column vector of jth column (j=1,2,3), hijIt is the H-matrix member of the i-th row jth column (i, j=1,2,3)
Element.Had according to the definition of homography matrix:
[h1 h2 h3]=K [r1 r2T] (4)
According to the property of spin matrix, r1And r2It is orthonormal vector, to have:
Wherein
It is respectively obtained by (5), (6)
h11h12·B11+(h31h12+h11h32)·B13+h21h22·B22+(h31h22+h21h32)·
B23+h31h32·B33=0 (8)
Camera production during, optical lens and imager chip coupling assembling when, it is desirable that the principal point of image from
Imager chip central point is without departing from certain pixel coverage.Therefore default principal point cxAnd cyInitial value be picture centre.(8),(9)
In formula, homography matrix H can be acquired by the calibrating template image of shooting, element hij(i, j=1,2,3) is known;cxWith
cyIt is it is known that B for image centerijIt is camera internal reference number focal length fx、fyWith principal point cx、cyMathematical combination, it was calculated
The intermediate quantity occurred in journey, BijIn only there are two unknown number fxAnd fy), so that two equation groups (8), (9) can solve two
Intrinsic parameter: focal length fxAnd fyInitial value.Spatial position initial value of the binocular camera in calibrating template coordinate system can basis
Camera is estimated relative to the position of calibrating template and the type of binocular camera.By taking mobile phone binocular camera mould group as an example,
Due to two camera optical axis less parallels, the initial value of relative rotation matrices R is set as unit matrix
Since camera lens is there are optical distortion, the pixel actually projected generally has a slightly offset on the image.Draw
That plays pattern distortion mainly has following reason: lens surface mismachining tolerance leads to its radial buckling existing defects;Each eyeglass
Optical centre cannot be stringent holding it is conllinear, produce eccentric error;Due to lens design, production and CCD camera assembling process
In tolerance, camera lens is not parallel with imager chip, there is inclination.The above error cause image simultaneously radially and tangentially generate it is abnormal
Become.Radial distortion refers to radially real image point on its ideal position and optical center connection, i.e., moves in or out.
Tangential distortion refer to real image point perpendicular to radial direction, i.e., it is tangential on, deviate.
The aforementioned theoretical location of pixels [x y] based on central projection model is shifted by distortion effects, practical to project
PositionIt is simulated with following relationship
Wherein, [k1k2k3] it is radial distortion parameter, [p1 p2] it is tangential distortion parameter, r2=x2+y2.It takes the photograph under normal circumstances
The distortion of the camera lens used as head is smaller, therefore radially and tangentially distortion parameter [k1k2k3] and [p1 p2] initial value it is general
It is set as zero.
4, the intrinsic parameter of binocular camera, outer parameter and distortion parameter are optimized, obtains accurate calibration result;
(corresponding to S4)
Each filled circles center of circle generates a corresponding image slices vegetarian refreshments in the picture when taking pictures.According to based on pinhole imaging system
The projection model of principle and above-mentioned distortion model, each center of circle can also calculate a theoretical imaging position and (produce comprising distortion
Raw offset).The deviation of real image point and the theory location of pixels is known as re-projection error.The ginseng of the geometric projection model
Number, the i.e. calibrating parameters of camera should make the quadratic sum of the re-projection error in all filled circles centers of circle on calibrating template minimum,
The projection model most accurately describes the optical imagery projection process of this depth of field depth camera at this time
Wherein M is the position in the filled circles center of circle on calibrating template, and m is that the centre point corresponds to actual pixels position in image,Attach most importance to projected position, i.e., the filled circles center of circle after central projection, because of figure
Location of pixels that image distortion and calculations of offset come out (AndDeviated comprising distortion), K is the intrinsic parameter of camera, rotating vector R
It is the outer parameter of camera, k=[k with translation vector T1k2k3] it is radial distortion parameter, p=[p1 p2] it is tangential distortion parameter.
Equation Levenberg-Marquardt, column Wen Baige-Ma Kuaerte algorithm optimizes, after several iteration, when repeatedly
When the error in generation is less than a preset threshold values, Optimized Iterative terminates, and the result K, R, T, k, p acquired is respectively the calibration distance
Corresponding intrinsic parameter, outer parameter and distortion parameter.The binocular camera calibrating parameters obtained by the nonlinear optimization are more accurate.
5, the relative position of binocular camera is determined;(corresponding to S5)
When binocular camera is demarcated, after left and right camera carries out the calibration of monocular camera respectively, two cameras
Position relative to approximately the same plane calibrating template is it is known that the spatial position of left camera is Rl、Tl, the space bit of right camera
It is set to Rr、Tr, it is so as to extrapolate the relative position of two cameras itself
6, depth information calculates
When binocular camera is respectively after four 0.5,1.0,1.5,2.0 meters of distance have been demarcated, be always obtained four groups it is double
The calibrating parameters of mesh camera.In depth test, VCM motor position when recording current bat figure, more current VCM motor position
VCM motor position corresponding with four groups of calibrating parameters selects the one group of calibrating parameters of VCM motor positional distance recently to calculate depth
Degree.
As shown in figure 5, the depth information calculates as follows: in left camera coordinate by taking one point P of object scene surface as an example
In system, optical center C1Coordinate is origin, and point P is in left camera imaging corresponding points m1Coordinate be [x1 y1f1].It is sat in right camera
In mark system, optical center C2Coordinate is origin, and point P is in right camera imaging corresponding points m2Coordinate be [x2 y2f2]。C1And m1On the right side
The coordinate of camera coordinate system is respectivelyAndP point coordinate is C1、m1Line and C2、m2Line
Intersection point under right camera coordinate system, depth information are the Z axis coordinate of the intersection point.
The prior art is compared, advantage and effect possessed by the present embodiment:
Advantage one: each calibration distance only uses a sheet of planar calibrating template, and the calibrating template does not need to rotate.With
The scaling method of the current general multiple angle bat figures of needs is compared, and is saved the producing line nominal time, is improved unit time production
Amount, reduces production cost.
Advantage two: binocular camera is demarcated in multiple calibration distances respectively, multiple groups binocular camera is always obtained
Calibrating parameters.Multiple groups calibrating parameters have effectively compensated for during automatic focusing camera head is taken pictures caused by lens location variation
Parameter differences can optimize the effect of camera 3D application to improve the precision of depth calculation, promote user experience, increase
Strong product competitiveness.
The parameter calibration method provided in the present embodiment can be applied to the camera (or camera) with automatic focusing function
Calibration, and due to carrying out parameter calibration only with a width calibrating template image, compared to existing general scaling method, there is side
Simple, the efficient advantage of method.In addition, the device provides the camera calibration at four distances, can better compensate for demarcating
Parameter difference because caused by changing lens location, so that successive depths calculating is more acurrate.
Method of the invention is equally applicable to the calibration of more mesh cameras, although there is multiple cameras, but as long as its depth
Computing Principle is carried out based on binocular principle, then is accordingly to be regarded as being binocular camera, belong to the scope of protection of the present invention within.
Claims (10)
1. a kind of auto-focusing binocular camera scaling method, it is characterised in that include the following steps:
A1, it obtains in multiple calibrating template images demarcated at distances at the respective distances of shooting, each calibration distance corresponding one
Individual a calibrating template image;Have size is known to demarcate pattern on the calibrating template;
This is calculated apart from institute apart from corresponding one individual calibrating template image in A2, analysis and each calibration of processing
Corresponding one group of binocular camera calibrating parameters;
A3, corresponding multiple calibration distances, are always obtained the calibrating parameters of multiple groups binocular camera.
2. auto-focusing binocular camera scaling method as described in claim 1, which is characterized in that in step Al, the mark
Determining pattern is solid circular array, which arranges a filled circles by C row × L and form, and C and L are natural number;Each filled circles
Size is identical, and radius is consistent, and both horizontally and vertically distance of center circle is all the same;The uniform cause of calibration pattern of each distance, by C
× L filled circles composition;The radius of circle and distance of center circle and binocular camera to the calibrating template of the calibration pattern of each distance
Apart from directly proportional.
3. auto-focusing binocular camera scaling method as claimed in claim 2, which is characterized in that in step Al, shooting
Two camera optical axises are each perpendicular to calibrating template when the calibrating template image or so so that the uncalibrated image focusing and not
In the presence of because of pattern deformation caused by shooting angle.
4. auto-focusing binocular camera scaling method as claimed in claim 2, which is characterized in that in step A2, analysis and
Handling each calibration, individual calibrating template image specifically comprises the following steps: apart from corresponding one
S2, in the calibrating template image that left and right camera is clapped, by image processing algorithm extract the filled circles center of circle where
Location of pixels;
S3, it is based on the calibrating template image, calculates the focal length of each camera lens of binocular camera, obtains each camera internal reference number, outer ginseng
The initial value of number, distortion parameter;
S4, it is based on Levenberg-Marquardt optimization algorithm, binocular camera intrinsic parameter, outer parameter, distortion parameter is carried out
Optimization, obtains accurate binocular camera calibrating parameters;
S5, calculate this apart from when the binocular camera relative position parameter.
5. auto-focusing binocular camera scaling method as claimed in claim 4, which is characterized in that in step S2, based on figure
As gray value detecting filled circles region, round homogeneous coordinates [x y 1] of the region mass center as the center of circle are sought.
6. auto-focusing binocular camera scaling method as claimed in claim 4, which is characterized in that in step S3, by such as
Lower equation group solves two intrinsic parameters, focal length fxAnd fy, initial value:
h11h12·B11+(h31h12+h11h32)·B13+h21h22·B22+(h31h22+h21h32)·B23+h31h32·B33=0
Wherein, hijIt is the homography matrix H-matrix element of the i-th row jth column (i, j=1,2,3), is known;
BijIt is determined by following formula:
In above formula, K is the intrinsic parameter of camera, is defined as
cxAnd cyIt is the principal point of image, presets principal point cxAnd cyInitial value be picture centre.
7. auto-focusing binocular camera scaling method as claimed in claim 4, which is characterized in that in step S4, when optimization,
So that the quadratic sum of the re-projection error in all filled circles centers of circle is minimum on calibrating template, it may be assumed that
Wherein M is the position in the filled circles center of circle on calibrating template, and m is that the centre point corresponds to actual pixels position in image,Attach most importance to projected position, i.e., the filled circles center of circle after central projection, because of figure
The location of pixels that image distortion and calculations of offset come out,AndComprising distortion deviate, K be camera intrinsic parameter, rotating vector R and
Translation vector T is the outer parameter of camera, k=[k1 k2 k3] it is radial distortion parameter, p=[p1 p2] it is tangential distortion parameter.
8. auto-focusing binocular camera scaling method as claimed in claim 4, which is characterized in that in step S5, two are taken the photograph
As the relative position of head itself is
Wherein, Rl、TlFor the spatial position of left camera, Rr、TrFor the spatial position of right camera;
If there is multiple cameras, then calculated according to this between any two.
9. a kind of auto-focusing binocular camera caliberating device, which is characterized in that using the calibration as described in claim 1 to 8
Method.
10. a kind of depth measurement method, which is characterized in that it is more to obtain correspondence using the scaling method as described in claim 1 to 8
The calibrating parameters of the multiple groups binocular camera of a calibration distance select when from current bat figure lens location most in depth test
A close calibration calculates depth apart from one group of corresponding binocular camera calibrating parameters.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811550433.9A CN109859272B (en) | 2018-12-18 | 2018-12-18 | Automatic focusing binocular camera calibration method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811550433.9A CN109859272B (en) | 2018-12-18 | 2018-12-18 | Automatic focusing binocular camera calibration method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109859272A true CN109859272A (en) | 2019-06-07 |
CN109859272B CN109859272B (en) | 2023-05-19 |
Family
ID=66891545
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811550433.9A Active CN109859272B (en) | 2018-12-18 | 2018-12-18 | Automatic focusing binocular camera calibration method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109859272B (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110345921A (en) * | 2019-06-12 | 2019-10-18 | 中国农业大学 | Stereoscopic fields of view vision measurement and vertical axial aberration and axial aberration bearing calibration and system |
CN110533728A (en) * | 2019-07-25 | 2019-12-03 | 长沙行深智能科技有限公司 | Scaling method, device and the medium of binocular stereo vision camera based on Zhang Zhengyou calibration |
CN111080705A (en) * | 2019-05-07 | 2020-04-28 | 像工场(深圳)科技有限公司 | Calibration method and device for automatic focusing binocular camera |
CN111314602A (en) * | 2020-02-17 | 2020-06-19 | 浙江大华技术股份有限公司 | Target object focusing method, target object focusing device, storage medium and electronic device |
CN111986512A (en) * | 2020-07-16 | 2020-11-24 | 华为技术有限公司 | Target distance determination method and device |
CN112862895A (en) * | 2019-11-27 | 2021-05-28 | 杭州海康威视数字技术股份有限公司 | Fisheye camera calibration method, device and system |
WO2022037633A1 (en) * | 2020-08-19 | 2022-02-24 | 展讯通信(上海)有限公司 | Calibration method and apparatus for binocular camera, image correction method and apparatus for binocular camera, storage medium, terminal and intelligent device |
CN114095657A (en) * | 2021-11-22 | 2022-02-25 | 成都天翼空间科技有限公司 | Automatic calibration method and system based on newly-added camera |
CN114205483A (en) * | 2022-02-17 | 2022-03-18 | 杭州思看科技有限公司 | Scanner precision calibration method and device and computer equipment |
CN115170674A (en) * | 2022-07-20 | 2022-10-11 | 禾多科技(北京)有限公司 | Camera principal point calibration method, device, equipment and medium based on single image |
CN115272491A (en) * | 2022-08-12 | 2022-11-01 | 哈尔滨工业大学 | Binocular PTZ camera dynamic self-calibration method |
CN117593225A (en) * | 2023-11-14 | 2024-02-23 | 自行科技(武汉)有限公司 | Method, system, equipment and storage medium for processing optical center deviation under electronic rearview mirror |
CN117593225B (en) * | 2023-11-14 | 2024-05-28 | 自行科技(武汉)有限公司 | Method, system, equipment and storage medium for processing optical center deviation under electronic rearview mirror |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101231750A (en) * | 2008-02-21 | 2008-07-30 | 南京航空航天大学 | Calibrating method of binocular three-dimensional measuring system |
CN101447073A (en) * | 2007-11-26 | 2009-06-03 | 新奥特(北京)视频技术有限公司 | Zoom lens calibration method |
CN103544699A (en) * | 2013-10-11 | 2014-01-29 | 江西省电力公司检修分公司 | Method for calibrating cameras on basis of single-picture three-circle template |
CN103559708A (en) * | 2013-10-30 | 2014-02-05 | 同济大学 | Industrial fixed-focus camera parameter calibration device based on square target model |
CN104155470A (en) * | 2014-07-15 | 2014-11-19 | 华南理工大学 | Detecting method and system based on binocular camera for real-time vehicle speed |
CN106131408A (en) * | 2016-07-11 | 2016-11-16 | 深圳市金立通信设备有限公司 | A kind of image processing method and terminal |
CN106846415A (en) * | 2017-01-24 | 2017-06-13 | 长沙全度影像科技有限公司 | A kind of multichannel fisheye camera binocular calibration device and method |
-
2018
- 2018-12-18 CN CN201811550433.9A patent/CN109859272B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101447073A (en) * | 2007-11-26 | 2009-06-03 | 新奥特(北京)视频技术有限公司 | Zoom lens calibration method |
CN101231750A (en) * | 2008-02-21 | 2008-07-30 | 南京航空航天大学 | Calibrating method of binocular three-dimensional measuring system |
CN103544699A (en) * | 2013-10-11 | 2014-01-29 | 江西省电力公司检修分公司 | Method for calibrating cameras on basis of single-picture three-circle template |
CN103559708A (en) * | 2013-10-30 | 2014-02-05 | 同济大学 | Industrial fixed-focus camera parameter calibration device based on square target model |
CN104155470A (en) * | 2014-07-15 | 2014-11-19 | 华南理工大学 | Detecting method and system based on binocular camera for real-time vehicle speed |
CN106131408A (en) * | 2016-07-11 | 2016-11-16 | 深圳市金立通信设备有限公司 | A kind of image processing method and terminal |
CN106846415A (en) * | 2017-01-24 | 2017-06-13 | 长沙全度影像科技有限公司 | A kind of multichannel fisheye camera binocular calibration device and method |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111080705A (en) * | 2019-05-07 | 2020-04-28 | 像工场(深圳)科技有限公司 | Calibration method and device for automatic focusing binocular camera |
CN111080705B (en) * | 2019-05-07 | 2021-06-04 | 像工场(深圳)科技有限公司 | Calibration method and device for automatic focusing binocular camera |
CN110345921A (en) * | 2019-06-12 | 2019-10-18 | 中国农业大学 | Stereoscopic fields of view vision measurement and vertical axial aberration and axial aberration bearing calibration and system |
CN110533728A (en) * | 2019-07-25 | 2019-12-03 | 长沙行深智能科技有限公司 | Scaling method, device and the medium of binocular stereo vision camera based on Zhang Zhengyou calibration |
CN112862895B (en) * | 2019-11-27 | 2023-10-10 | 杭州海康威视数字技术股份有限公司 | Fisheye camera calibration method, device and system |
CN112862895A (en) * | 2019-11-27 | 2021-05-28 | 杭州海康威视数字技术股份有限公司 | Fisheye camera calibration method, device and system |
CN111314602A (en) * | 2020-02-17 | 2020-06-19 | 浙江大华技术股份有限公司 | Target object focusing method, target object focusing device, storage medium and electronic device |
CN111314602B (en) * | 2020-02-17 | 2021-09-17 | 浙江大华技术股份有限公司 | Target object focusing method, target object focusing device, storage medium and electronic device |
CN111986512A (en) * | 2020-07-16 | 2020-11-24 | 华为技术有限公司 | Target distance determination method and device |
WO2022037633A1 (en) * | 2020-08-19 | 2022-02-24 | 展讯通信(上海)有限公司 | Calibration method and apparatus for binocular camera, image correction method and apparatus for binocular camera, storage medium, terminal and intelligent device |
CN114095657A (en) * | 2021-11-22 | 2022-02-25 | 成都天翼空间科技有限公司 | Automatic calibration method and system based on newly-added camera |
CN114095657B (en) * | 2021-11-22 | 2024-02-27 | 成都天翼空间科技有限公司 | Automatic calibration method and system based on newly added camera |
CN114205483A (en) * | 2022-02-17 | 2022-03-18 | 杭州思看科技有限公司 | Scanner precision calibration method and device and computer equipment |
CN114205483B (en) * | 2022-02-17 | 2022-07-29 | 杭州思看科技有限公司 | Scanner precision calibration method and device and computer equipment |
CN115170674A (en) * | 2022-07-20 | 2022-10-11 | 禾多科技(北京)有限公司 | Camera principal point calibration method, device, equipment and medium based on single image |
CN115272491A (en) * | 2022-08-12 | 2022-11-01 | 哈尔滨工业大学 | Binocular PTZ camera dynamic self-calibration method |
CN117593225A (en) * | 2023-11-14 | 2024-02-23 | 自行科技(武汉)有限公司 | Method, system, equipment and storage medium for processing optical center deviation under electronic rearview mirror |
CN117593225B (en) * | 2023-11-14 | 2024-05-28 | 自行科技(武汉)有限公司 | Method, system, equipment and storage medium for processing optical center deviation under electronic rearview mirror |
Also Published As
Publication number | Publication date |
---|---|
CN109859272B (en) | 2023-05-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109859272A (en) | A kind of auto-focusing binocular camera scaling method and device | |
CN109767476A (en) | A kind of calibration of auto-focusing binocular camera and depth computing method | |
CN110285793B (en) | Intelligent vehicle track measuring method based on binocular stereo vision system | |
CN110322702B (en) | Intelligent vehicle speed measuring method based on binocular stereo vision system | |
CN106091984B (en) | A kind of three dimensional point cloud acquisition methods based on line laser | |
CN105758426B (en) | The combined calibrating method of the multisensor of mobile robot | |
CN106408556B (en) | A kind of small items measuring system scaling method based on general imaging model | |
CN105716542B (en) | A kind of three-dimensional data joining method based on flexible characteristic point | |
CN104616292B (en) | Monocular vision measuring method based on global homography matrix | |
CN104173054B (en) | Measuring method and measuring device for height of human body based on binocular vision technique | |
CN102376089B (en) | Target correction method and system | |
CN104537707B (en) | Image space type stereoscopic vision moves real-time measurement system online | |
CN110728715A (en) | Camera angle self-adaptive adjusting method of intelligent inspection robot | |
CN110044300A (en) | Amphibious 3D vision detection device and detection method based on laser | |
CN108844459A (en) | A kind of scaling method and device of leaf digital template detection system | |
CN109579695B (en) | Part measuring method based on heterogeneous stereoscopic vision | |
CN104036542B (en) | Spatial light clustering-based image surface feature point matching method | |
CN109615661A (en) | Light-field camera intrinsic parameter caliberating device and method | |
CN101231750A (en) | Calibrating method of binocular three-dimensional measuring system | |
CN109523595A (en) | A kind of architectural engineering straight line corner angle spacing vision measuring method | |
CN112161997B (en) | Online precise visual measurement method and system for three-dimensional geometric dimension of semiconductor chip pin | |
CN107941153B (en) | Visual system for optimizing calibration of laser ranging | |
CN111080705B (en) | Calibration method and device for automatic focusing binocular camera | |
CN109712232B (en) | Object surface contour three-dimensional imaging method based on light field | |
CN109272555B (en) | External parameter obtaining and calibrating method for RGB-D camera |
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 |