CN107170043B - A kind of three-dimensional rebuilding method - Google Patents
A kind of three-dimensional rebuilding method Download PDFInfo
- Publication number
- CN107170043B CN107170043B CN201710465790.4A CN201710465790A CN107170043B CN 107170043 B CN107170043 B CN 107170043B CN 201710465790 A CN201710465790 A CN 201710465790A CN 107170043 B CN107170043 B CN 107170043B
- Authority
- CN
- China
- Prior art keywords
- cosθ
- point
- rotating platform
- camera
- dimensional
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
Abstract
The invention belongs to three-dimensional reconstruction fields, are related to a kind of three-dimensional rebuilding method.Present invention incorporates technique of binocular stereoscopic vision, structured light technique and rotating platform technologies, propose the three-dimensional rebuilding method that a kind of precision is high, speed is fast, at low cost;Possess compared to binocular stereo vision and rebuild speed faster, precision is higher, and regional effect not abundant for article surface vein is good.Compared to traditional structured light technique, this system is not necessarily to the trouble of projector calibrating, replaces projector with second camera, and the calibration of camera is more more accurate than the calibration of projector, so the three-dimensional point cloud rebuild is also better.The multi-angle of view point cloud of use uses the stitching algorithm of rotating platform, and compared to traditional ICP method for registering, this method point cloud is faster, easy to operate to participate in without artificial woth no need to initial point cloud data.
Description
Technical field
The invention belongs to three-dimensional reconstruction fields, are related to a kind of three-dimensional rebuilding method.
Background technique
Three-dimensional reconstruction is a branch of computer vision, is computer vision and computer graphic image processing phase
In conjunction with a research direction.It is widely used in industrial automation, reverse-engineering, historical relic's protection, computer aided medicine, void
The scenes such as quasi- reality, augmented reality and robot application.
Technique of binocular stereoscopic vision and structured light technique are two kinds of typical three-dimensional reconstructions, are respectively had the advantage that and not
Foot.Technique of binocular stereoscopic vision is the left and right two images by camera in two angle shot objects, then utilizes three-dimensional
Same place in the two images of left and right, and combining camera inside and outside parameter are searched out with algorithm, goes out measured object using triangle intersection calculation
The three-dimensional space position coordinate information of body.Technique of binocular stereoscopic vision does not need actively to project pattern-information, and hardware configuration is simple,
But reconstruction point cloud precision is low, rebuilds slow, few for the surface texture information object of speed, is easy to appear match point error etc. and lacks
Point.Structured light technique is to project specific coding pattern to body surface by projector, then shot by body surface by camera
The coding pattern modulated recovers the depth information of object by the decoding operation of coding pattern.Structured light technique rebuilds essence
Degree is high, and speed is fast, even the few object of surface texture information can also obtain rebuilding effect well, but traditional structure light
Reconstructing system is all monocular mostly, needs to demarcate projector during calculating depth information, and projector
Calibration process is again extremely cumbersome.
Summary of the invention
To be solved by this invention to be, aiming at above-mentioned conventional method, there are problems, propose a kind of combination binocular solid
The three-dimensional rebuilding method of vision, structure light and rotating platform.
For the ease of the understanding to the technology of the present invention, the geometrical model that the present invention uses is introduced first, as shown in Figure 1, its
In, Ol is left camera photocentre, and Or is right camera photocentre, and P is any point in space, and Pl and Pr are respectively P point in left and right camera
In picture point, referred to as a pair of of same place, plane POlOr and left and right are known as a pair of of polar curve as intersection Lpl, Lpr of plane.
The technical scheme is that as shown in Fig. 2, a kind of three-dimensional rebuilding method, which is characterized in that including following step
It is rapid:
S1, camera parameter calibration, the relative position square between the inside and outside parameter and two cameras of two cameras in calibration left and right
Battle array;According to practical situations, the gridiron pattern calibration for cameras internal reference of Zhang Zhengyou and the spin matrix R peace of two cameras can be used
Move matrix T
S2, the three-dimensional coordinate point for obtaining the single angle of target object, as shown in figure 3, specifically including:
S21, six gray encoding images are projected to target object, after the camera shooting of left and right two, to shooting
Image is decoded, then each pixel has Gray's code value in image;
S22,4 sinusoidal phase-shift coding images are projected to object, after the camera shooting of left and right two, to the image of shooting
It is decoded, then each pixel has a phase value in image;
S23, by the method for sub-pix homotopy mapping in left images, for the certain point p in left imagel(ul,
vl), its homonymy matching point in right image is found, is specifically included:
S231, according to the relative position matrix between the intrinsic parameter and camera of camera, find out point pl(ul,vl) in right image
In polar curve l;
S232, Gray's code value and point p on polar curve l are calculatedl(ul,vl) at the equal region A of Gray's code value;
S233, linear fit operation is carried out to the sinusoidal phase value of pixel in the A of region, obtains the straight line of linear fit;
S234, it is found out on the straight line of linear fit and point pl(ul,vl) at the equal u axial coordinate u of phaser, then by urIt sits
Marker tape enters polar curve l and obtains v axial coordinate vr, point p at this timer(ur,vr) it is pl(ul,vl) corresponding sub-pix is same in right image
Famous cake;
S24, according to step S23 obtain homonymy matching point and camera inside and outside parameter, in conjunction with binocular stereo vision be imaged
Geometrical model obtains the three-dimensional coordinate point of the single angle of target object;
The purpose of above scheme is to find sub-pix same place, utilizes phase value periodic line after the decoding of sinusoidal phase shift striped
Property variation characteristic combination Gray code pattern and photography geometry in limiting constraint, to phase value progress linear fit, obtain
The same place for having arrived left images Central Asia pixel matching, compared to the matched same place of pixel precision, the object being finally calculated
Body three-dimensional point cloud, surface is more smooth, and point cloud distribution is more uniform, and error is also smaller.
S3, the complexity according to actual object, multiple rotary rotating platform repeat step S2 to each angle, obtain
To the three-dimensional coordinate point of the multiple angles of target object.
S4: using step S2, individually rebuild rotating platform, carries out plane fitting to rotating platform, the specific steps are as follows:
S41, the following formula 1 of Plane Equation for constructing rotating platform:
Ax+by+cz+d=0 (formula 1)
S42, the n point cloud data (x by rotating platformi,yi,zi) substitute into formula 1 can obtain following formula 2:
S43, a, b, c, the value of d, the then unit normal vector of rotating platform plane are obtained by least square solution formula 2
For following formula 3:
Then the center O of rotating platform plane is following formula 4:
S5, spliced using point cloud of the rotating platform to n angle of target object, specifically:
S51, building rotating platform splice the following formula 5 of multi-angle point cloud model:
Wherein shown in the following formula 6 of R:
In formula 6:
R1=NxNy(1-cosθ)+cosθ
R2=NxNy(1-cosθ)+Nz sinθ
R3=NxNz(1-cosθ)-Ny sinθ
R4=NxNy(1-cosθ)-Nz sinθ
R5=NyNy(1-cosθ)+cosθ
R6=NyNz(1-cosθ)+Nx sinθ
R7=NxNz(1-cosθ)+Ny sinθ
R8=NyNz(1-cosθ)-Nx sinθ
R9=NzNz(1-cosθ)+cosθ
Wherein, O (Xc,Yc,Zc) it is coordinate of the center of rotating platform plane under reference frame, N (Nx,Ny,Nz) be
Unit vector of the plane normal of rotating platform under reference frame, the O and N solved in as step S4, θ are two panels point
Rotation angle between cloud;
S52, using the formula 5 in S5, the three-dimensional coordinate point of the multiple angles of the target object generated in S3 is spliced to together
Under one coordinate system
S6, target complete three-dimensional model is obtained.
The world coordinate system of three-dimensional reconstruction of the present invention is all the camera coordinates system of left camera, using this feature, to rotation
Platform carries out three-dimensional reconstruction, then carries out plane fitting to the point cloud of rotating platform, then the normal of plane is exactly rotating platform
Coordinate of the normal under world coordinate system, the center of gravity for putting cloud is exactly coordinate of the center of rotating platform under world coordinate system.This
For kind calibration algorithm without specific calibration object, algorithm is simple and convenient
The invention has the benefit that present invention incorporates technique of binocular stereoscopic vision, structured light technique and rotating platforms
Technology proposes the three-dimensional rebuilding method that a kind of precision is high, speed is fast, at low cost;Possess reconstruction compared to binocular stereo vision
Faster, precision is higher for speed, and regional effect not abundant for article surface vein is good.Compared to traditional structure light
Technology, this system are not necessarily to the trouble of projector calibrating, replace projector with second camera, and the calibration of camera is than projector
Calibration it is more accurate, so rebuild three-dimensional point cloud it is also better.The multi-angle of view point cloud of use is using rotating platform
Stitching algorithm, compared to traditional ICP method for registering, faster, woth no need to initial point cloud data, operation is simple for this method point cloud
It is single to be participated in without artificial.
Detailed description of the invention
Fig. 1 is the three-dimensional reconstruction geometrical model that the present invention uses;
Fig. 2 is three-dimensional reconstruction flow chart;
Fig. 3 is the three-dimensional reconstruction flow chart of one angle of object.
Specific embodiment
Summary is described in detail technical solution of the present invention, and details are not described herein.
Claims (1)
1. a kind of three-dimensional rebuilding method, which comprises the following steps:
S1, camera parameter calibration, the relative position matrix between the inside and outside parameter and two cameras of two cameras in calibration left and right;
S2, the three-dimensional coordinate point for obtaining the single angle of target object, specifically include:
S21, six gray encoding images are projected to target object, after the camera shooting of left and right two, to the image of shooting
It is decoded, then each pixel has Gray's code value in image;
S22,4 sinusoidal phase-shift coding images are projected to object, after the camera shooting of left and right two, the image of shooting is carried out
Decoding, then each pixel has a phase value in image;
S23, by the method for sub-pix homotopy mapping in left images, for the certain point p in left imagel(ul,vl), it seeks
Its homonymy matching point in right image is looked for, is specifically included:
S231, according to the relative position matrix between the intrinsic parameter and camera of camera, find out point pl(ul,vl) in right image
Polar curve l;
S232, Gray's code value and point p on polar curve l are calculatedl(ul,vl) at the equal region A of Gray's code value;
S233, linear fit operation is carried out to the sinusoidal phase value of pixel in the A of region, obtains the straight line of linear fit;
S234, it is found out on the straight line of linear fit and point pl(ul,vl) at the equal u axial coordinate u of phaser, then by urGrid zone
Enter polar curve l and obtains v axial coordinate vr, point p at this timer(ur,vr) it is pl(ul,vl) corresponding sub-pix is of the same name in right image
Point;
S24, according to step S23 obtain homonymy matching point and camera inside and outside parameter, in conjunction with binocular stereo vision imaging geometry
Model obtains the three-dimensional coordinate point of the single angle of target object;
S3, the complexity according to actual object, multiple rotary rotating platform repeat step S2 to each angle, obtain mesh
Mark the three-dimensional coordinate point of the multiple angles of object;
S4: using step S2, individually rebuild rotating platform, carries out plane fitting to rotating platform, the specific steps are as follows:
S41, the following formula 1 of Plane Equation for constructing rotating platform:
Ax+by+cz+d=0 (formula 1)
S42, the n point cloud data (x by rotating platformi,yi,zi) substitute into formula 1 can obtain following formula 2:
S43, a, b, c are obtained by least square solution formula 2, the value of d, then the unit normal vector of rotating platform plane is such as
Lower formula 3:
Then the center O of rotating platform plane is following formula 4:
S5, spliced using point cloud of the rotating platform to n angle of target object, specifically:
S51, building rotating platform splice the following formula 5 of multi-angle point cloud model:
Wherein shown in the following formula 6 of R:
In formula 6:
R1=NxNy(1-cosθ)+cosθ
R2=NxNy(1-cosθ)+Nzsinθ
R3=NxNz(1-cosθ)-Nysinθ
R4=NxNy(1-cosθ)-Nzsinθ
R5=NyNy(1-cosθ)+cosθ
R6=NyNz(1-cosθ)+Nxsinθ
R7=NxNz(1-cosθ)+Nysinθ
R8=NyNz(1-cosθ)-Nxsinθ
R9=NzNz(1-cosθ)+cosθ
Wherein, O (Xc,Yc,Zc) it is coordinate of the center of rotating platform plane under reference frame, N (Nx,Ny,Nz) it is rotation
Unit vector of the plane normal of platform under reference frame, the O and N solved in as step S4, θ be two panels point cloud it
Between rotation angle;
S52, using the formula 5 in S51, the three-dimensional coordinate point of the multiple angles of the target object generated in S3 is spliced to same
Under coordinate system;
S6, target complete three-dimensional model is obtained.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710465790.4A CN107170043B (en) | 2017-06-19 | 2017-06-19 | A kind of three-dimensional rebuilding method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710465790.4A CN107170043B (en) | 2017-06-19 | 2017-06-19 | A kind of three-dimensional rebuilding method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107170043A CN107170043A (en) | 2017-09-15 |
CN107170043B true CN107170043B (en) | 2019-06-18 |
Family
ID=59819842
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710465790.4A Expired - Fee Related CN107170043B (en) | 2017-06-19 | 2017-06-19 | A kind of three-dimensional rebuilding method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107170043B (en) |
Families Citing this family (28)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107918948B (en) * | 2017-11-02 | 2021-04-16 | 深圳市自由视像科技有限公司 | 4D video rendering method |
CN108288292A (en) * | 2017-12-26 | 2018-07-17 | 中国科学院深圳先进技术研究院 | A kind of three-dimensional rebuilding method, device and equipment |
CN108335350A (en) * | 2018-02-06 | 2018-07-27 | 聊城大学 | The three-dimensional rebuilding method of binocular stereo vision |
CN108665535A (en) * | 2018-05-10 | 2018-10-16 | 青岛小优智能科技有限公司 | A kind of three-dimensional structure method for reconstructing and system based on coding grating structured light |
CN108895969A (en) * | 2018-05-23 | 2018-11-27 | 深圳大学 | A kind of 3 D detection method and device of phone housing |
CN108961378B (en) * | 2018-07-05 | 2023-03-28 | 深圳辰视智能科技有限公司 | Multi-eye point cloud three-dimensional reconstruction method, device and equipment |
CN109064536B (en) * | 2018-07-27 | 2022-12-06 | 电子科技大学 | Page three-dimensional reconstruction method based on binocular structured light |
CN109648202B (en) * | 2018-08-20 | 2021-02-12 | 南京理工大学 | Additive manufacturing forming precision control method for non-flat surface autonomous recognition robot |
CN109253706B (en) * | 2018-08-24 | 2020-07-03 | 中国科学技术大学 | Tunnel three-dimensional topography measuring method based on digital image |
CN109544454B (en) * | 2018-11-20 | 2023-03-31 | 广州方舆科技有限公司 | Automatic splicing method of point cloud data |
CN109541186B (en) * | 2018-11-29 | 2021-11-09 | 烟台大学 | Coarse aggregate compactness calculation method based on shape parameters |
CN109499010B (en) * | 2018-12-21 | 2021-06-08 | 苏州雷泰医疗科技有限公司 | Radiotherapy auxiliary system based on infrared and visible light three-dimensional reconstruction and method thereof |
CN109712200B (en) * | 2019-01-10 | 2023-03-14 | 深圳大学 | Binocular positioning method and system based on least square principle and side length reckoning |
CN110288581B (en) * | 2019-06-26 | 2022-11-04 | 电子科技大学 | Segmentation method based on model for keeping shape convexity level set |
CN110288699A (en) * | 2019-06-26 | 2019-09-27 | 电子科技大学 | A kind of three-dimensional rebuilding method based on structure light |
CN110349251B (en) * | 2019-06-28 | 2020-06-16 | 深圳数位传媒科技有限公司 | Three-dimensional reconstruction method and device based on binocular camera |
CN110414435A (en) * | 2019-07-29 | 2019-11-05 | 深兰科技(上海)有限公司 | The generation method and equipment of three-dimensional face data based on deep learning and structure light |
CN110926371A (en) * | 2019-11-19 | 2020-03-27 | 宁波舜宇仪器有限公司 | Three-dimensional surface detection method and device |
CN110910506B (en) * | 2019-12-03 | 2021-02-09 | 江苏集萃华科智能装备科技有限公司 | Three-dimensional reconstruction method and device based on normal detection, detection device and system |
CN111127633A (en) * | 2019-12-20 | 2020-05-08 | 支付宝(杭州)信息技术有限公司 | Three-dimensional reconstruction method, apparatus, and computer-readable medium |
CN111063016A (en) * | 2019-12-31 | 2020-04-24 | 螳螂慧视科技有限公司 | Multi-depth lens face modeling method and system, storage medium and terminal |
CN111275776A (en) * | 2020-02-11 | 2020-06-12 | 北京淳中科技股份有限公司 | Projection augmented reality method and device and electronic equipment |
CN111462331B (en) * | 2020-03-31 | 2023-06-27 | 四川大学 | Lookup table method for expanding epipolar geometry and calculating three-dimensional point cloud in real time |
CN112509138B (en) * | 2020-12-11 | 2023-07-18 | 电子科技大学 | LCOS-based high-precision three-dimensional reconstruction system for indoor plastering robot |
CN112967398B (en) * | 2021-03-01 | 2023-07-25 | 北京奇艺世纪科技有限公司 | Three-dimensional data reconstruction method and device and electronic equipment |
CN113074661B (en) * | 2021-03-26 | 2022-02-18 | 华中科技大学 | Projector corresponding point high-precision matching method based on polar line sampling and application thereof |
CN113362447B (en) * | 2021-05-25 | 2023-04-07 | 天津大学 | Surface normal reconstruction fusion system and reconstruction fusion method |
CN113532325B (en) * | 2021-06-08 | 2023-05-05 | 深圳市格灵精睿视觉有限公司 | Dynamic step number phase decomposition method, electronic equipment and computer readable storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101697233A (en) * | 2009-10-16 | 2010-04-21 | 长春理工大学 | Structured light-based three-dimensional object surface reconstruction method |
CN102663753A (en) * | 2012-04-12 | 2012-09-12 | 浙江大学 | Body surface rebuilding method and body surface rebuilding device based on multiple visual angles of binocular stereoscopic vision |
CN103759670A (en) * | 2014-01-06 | 2014-04-30 | 四川虹微技术有限公司 | Object three-dimensional information acquisition method based on digital close range photography |
-
2017
- 2017-06-19 CN CN201710465790.4A patent/CN107170043B/en not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101697233A (en) * | 2009-10-16 | 2010-04-21 | 长春理工大学 | Structured light-based three-dimensional object surface reconstruction method |
CN102663753A (en) * | 2012-04-12 | 2012-09-12 | 浙江大学 | Body surface rebuilding method and body surface rebuilding device based on multiple visual angles of binocular stereoscopic vision |
CN103759670A (en) * | 2014-01-06 | 2014-04-30 | 四川虹微技术有限公司 | Object three-dimensional information acquisition method based on digital close range photography |
Also Published As
Publication number | Publication date |
---|---|
CN107170043A (en) | 2017-09-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107170043B (en) | A kind of three-dimensional rebuilding method | |
Debevec et al. | Modeling and rendering architecture from photographs: A hybrid geometry-and image-based approach | |
CN110288642B (en) | Three-dimensional object rapid reconstruction method based on camera array | |
CN108288292A (en) | A kind of three-dimensional rebuilding method, device and equipment | |
CN104240289B (en) | Three-dimensional digitalization reconstruction method and system based on single camera | |
Bonfort et al. | General specular surface triangulation | |
WO2017008226A1 (en) | Three-dimensional facial reconstruction method and system | |
JP2019516983A (en) | Three-dimensional scanning system and scanning method thereof | |
CN107358633A (en) | Join scaling method inside and outside a kind of polyphaser based on 3 points of demarcation things | |
CN102032878A (en) | Accurate on-line measurement method based on binocular stereo vision measurement system | |
CN106504321A (en) | Method using the method for photo or video reconstruction three-dimensional tooth mould and using RGBD image reconstructions three-dimensional tooth mould | |
CN111091599B (en) | Multi-camera-projector system calibration method based on sphere calibration object | |
CN114998499A (en) | Binocular three-dimensional reconstruction method and system based on line laser galvanometer scanning | |
CN108010125A (en) | True scale three-dimensional reconstruction system and method based on line-structured light and image information | |
Pagani et al. | Dense 3D Point Cloud Generation from Multiple High-resolution Spherical Images. | |
CN108876909A (en) | A kind of three-dimensional rebuilding method based on more image mosaics | |
CN110425983A (en) | A kind of monocular vision three-dimensional reconstruction distance measuring method based on polarization multi-spectrum | |
CN113129430A (en) | Underwater three-dimensional reconstruction method based on binocular structured light | |
Zhang et al. | Ray space features for plenoptic structure-from-motion | |
CN106683163A (en) | Imaging method and system used in video monitoring | |
Hafeez et al. | Image based 3D reconstruction of texture-less objects for VR contents | |
Xiao et al. | 3D integral imaging using sparse sensors with unknown positions | |
CN102269575A (en) | Mapping-based phase matching method in vision measurement | |
Fleischmann et al. | Fast projector-camera calibration for interactive projection mapping | |
Hafeez et al. | 3D surface reconstruction of smooth and textureless objects |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190618 Termination date: 20200619 |
|
CF01 | Termination of patent right due to non-payment of annual fee |