CN107121131A - A kind of horizontal relative pose recognition methods of binocular camera - Google Patents
A kind of horizontal relative pose recognition methods of binocular camera Download PDFInfo
- Publication number
- CN107121131A CN107121131A CN201710219009.5A CN201710219009A CN107121131A CN 107121131 A CN107121131 A CN 107121131A CN 201710219009 A CN201710219009 A CN 201710219009A CN 107121131 A CN107121131 A CN 107121131A
- Authority
- CN
- China
- Prior art keywords
- thing
- camera
- hand edge
- identification
- horizontal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C11/00—Photogrammetry or videogrammetry, e.g. stereogrammetry; Photographic surveying
- G01C11/02—Picture taking arrangements specially adapted for photogrammetry or photographic surveying, e.g. controlling overlapping of pictures
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C11/00—Photogrammetry or videogrammetry, e.g. stereogrammetry; Photographic surveying
- G01C11/04—Interpretation of pictures
- G01C11/30—Interpretation of pictures by triangulation
Abstract
The present invention relates to technical field of computer vision, a kind of horizontal relative pose recognition methods of binocular camera comprises the following steps：(1) identification thing and binocular camera are installed, (2) binocular image of collection identification thing, (3) plane coordinate system is built, (4) angle information of the identification thing in binocular camera image is obtained, (5) distance at identification thing edge and camera photocentre is calculated, (6) calculate level angle information of the identification thing with respect to binocular camera.The present invention uses Computer Vision Recognition posture, and equipment cost is low, it is not necessary to be confined in fixed scene, system is still effective when robot location moves.In addition, the inventive method is feasible simple, complexity is low.
Description
Technical field
The present invention relates to a kind of horizontal relative pose recognition methods of binocular camera, belong to technical field of computer vision.
Background technology
The vision of the mankind can not only recognize the external performances such as object color shape, can also be by contrasting two eyes institutes
The range of information such as depth, the corner posture that the difference of acquisition image is analyzed to obtain image.Binocular vision is that machine is regarded
The important form felt, is also the developing direction of the machine vision of following main flow.Binocular vision is by using apery class
Visual manner, according to the principle of camera pin-hole imaging model, obtains the information of image, and lead to by the camera of left and right two
Cross the image difference for calculating different visual angles, the corresponding relation of binding characteristic point, to obtain depth, the corner of target, the information such as posture,
So as to provide more data signals for machine vision.
Currently for multiple target pose measurement, mainly regarded using microwave radar, laser technology, Position and attitude sensor and computer
The technologies such as feel.Microwave radar technology is mainly used in large-range measuring, and measurement accuracy is low and expensive；Laser technology is obtained
Point cloud information correction after precision it is high, but the process of correction and process points cloud is very complicated, and three-dimensional laser equipment price is expensive；
Position and attitude sensor technology is easy to use, but is only applicable to mini-system, and due to the transmission speed problem of radio communication, institute
There is measurement delay, measurement real-time is bad；The application of computer vision also very extensively, including panoramic vision, monocular vision,
Binocular vision etc., wherein, panoramic vision contains much information, pattern distortion is larger, causes algorithm complicated, the shortcoming more than garbage；
Monocular vision is simple in construction, but it is few to there is information content, the problem of obtaining not enough to spatial information；Binocular vision passes through to double
The image difference of mesh camera carries out analysis identification pose, has the advantages that cheap, information is comprehensive, but there is also image
The problem of handling relative complex.It is therefore proposed that the small binocular vision recognition methods of a kind of method simple possible, amount of calculation is urgently
The problem of solution.
The content of the invention
In order to overcome the deficiencies in the prior art, it is an object of the present invention to provide a kind of binocular camera level with respect to position
Appearance recognition methods.This method simple possible, the characteristics of take full advantage of binocular vision and advantage, by carrying out geometry to image difference
Relational calculus just can obtain horizontal attitude corner.
In order to realize foregoing invention purpose, problem present in prior art is solved, the technical solution adopted by the present invention is：
A kind of horizontal relative pose recognition methods of binocular camera, comprises the following steps：
Step 1, installation identification thing and binocular camera, fix a knowledge on the object for needing the horizontal relative pose of identification
Other thing, identification thing needs the condition met to be to have two linear edges perpendicular to the ground, by binocular camera horizontal positioned, then
Identification thing is placed in the image pickup scope of binocular camera, to ensure the image information that can collect identification thing；
The binocular image of step 2, collection identification thing, opens operation camera, is gathered out by binocular camera with for the moment
Carve the image of two different azimuths to recognizing thing；
Step 3, build plane coordinate system, cross binocular camera left and right camera photocentre where straight line build parallel to
The plane coordinate system (x, y) on ground, using the straight line where the camera photocentre line of left and right as x-axis, left camera photocentre points to right
Camera photocentre is x-axis positive direction；Using excessively left camera photocentre and perpendicular to x-axis horizontal direction as y-axis, point to identification thing one
The direction of side is y-axis positive direction；Here, also it is left camera photocentre, O if O points are coordinate origin1Point is right camera light
The heart, line segment AB is identification object plane (plane perpendicular to ground is shown as line segment in a top view)；A points and B points are respectively to know
Other thing left hand edge and right hand edge (line segment perpendicular to ground is shown as a little in a top view)；
The angle information of step 4, acquisition identification thing in binocular camera image, obtains in binocular camera image and recognizes
The angle of thing in the horizontal direction, including the horizontal sextant angle ∠ between thing left hand edge and right hand edge is recognized in left camera image
AOB is set to θ1, identification thing right hand edge and the positive direction horizontal sextant angle ∠ BOO of x-axis in left camera image1It is set to θ2, right camera
The horizontal sextant angle ∠ AO between thing left hand edge and right hand edge are recognized in image1B is set to θ3, the identification thing left side in right camera image
Edge and the horizontal sextant angle ∠ AO of x-axis negative direction1O is set to θ4；
Step 5, the distance for calculating identification thing edge and camera photocentre, with left camera photocentre O, right camera photocentre O1
The triangle Δ OO surrounded in the horizontal direction with identification thing left hand edge A1A, is obtained according to triangle sine,
In formula, l1Represent right camera photocentre O1Horizontal range with recognizing thing left hand edge A, i.e. line segment AO1Length, d tables
Show the distance of camera photocentre in left and right in binocular camera, i.e. line segment OO1Length, right camera photocentre O is solved by formula (1)1
Horizontal range l with recognizing thing left hand edge A1For,
Similarly, with left camera photocentre O, right camera photocentre O1Surrounded in the horizontal direction with identification thing right hand edge B
Triangle Δ OO1B, is obtained according to triangle sine,
In formula, l2Represent right camera photocentre O1Horizontal range with recognizing thing right hand edge B, i.e. line segment BO1Length, by
Formula (3) solves right camera photocentre O1Horizontal range l with recognizing thing right hand edge B2For,
Step 6, calculating recognize level angle information of the thing with respect to binocular camera, with right camera photocentre O1, identification thing
Left hand edge A is with recognizing thing right hand edge B surrounded triangle Δ AO in the horizontal direction1B, is obtained according to the triangle cosine law,
l2=l1 2+l2 2-2l1·l2·cosθ3 (5)
In formula, l represents to recognize the length of thing left hand edge A and right hand edge B horizontal range, i.e. line segment AB, is solved by formula (5)
The horizontal range l that thing left hand edge A and right hand edge B must be recognized is,
Further according to triangle Δ AO1Sine is obtained in B,
In formula, θ5Represent line segment AB and line segment BO1The horizontal sextant angle ∠ ABO formed1, solved by formula (7),
Further according to triangle geometrical relationship, try to achieve identification thing is with respect to the level angle θ of x-axis,
Present invention has the advantages that：A kind of horizontal relative pose recognition methods of binocular camera, comprises the following steps：(1)
Identification thing and binocular camera are installed, (2) collection recognizes the binocular image of thing, and (3) build plane coordinate system, and (4) obtain identification
Angle information of the thing in binocular camera image, (5) calculate the distance at identification thing edge and camera photocentre, and (6), which are calculated, knows
Other thing is with respect to the level angle information of binocular camera, and compared with the prior art, the present invention uses Computer Vision Recognition posture,
Equipment cost is low, it is not necessary to be confined in fixed scene, and system is still effective when robot location moves.In addition, of the invention
Method is feasible simple, and complexity is low.
Brief description of the drawings
Fig. 1 is the inventive method flow chart of steps.
Fig. 2 is the 3 d effect graph of pose identifying system of the present invention.
In figure：(a) be pose identifying system of the present invention front view.
(b) be pose identifying system of the present invention top view.
(c) be pose identifying system of the present invention 3-D view.
Fig. 3 is pose recognition principle schematic diagram in top view of the present invention.
Embodiment
The invention will be further described below in conjunction with the accompanying drawings.
As shown in figure 1, a kind of horizontal relative pose recognition methods of binocular camera, it is characterised in that comprise the following steps：
Step 1, installation identification thing and binocular camera, fix a knowledge on the object for needing the horizontal relative pose of identification
Other thing, identification thing needs the condition met to be to have two linear edges perpendicular to the ground, by binocular camera horizontal positioned, then
Identification thing is placed in the image pickup scope of binocular camera, to ensure the image information that can collect identification thing；
The binocular image of step 2, collection identification thing, opens operation camera, is gathered out by binocular camera with for the moment
Carve the image of two different azimuths to recognizing thing；
Step 3, build plane coordinate system, cross binocular camera left and right camera photocentre where straight line build parallel to
The plane coordinate system (x, y) on ground, using the straight line where the camera photocentre line of left and right as x-axis, left camera photocentre points to right
Camera photocentre is x-axis positive direction；Using excessively left camera photocentre and perpendicular to x-axis horizontal direction as y-axis, point to identification thing one
The direction of side is y-axis positive direction；Here, also it is left camera photocentre, O if O points are coordinate origin1Point is right camera light
The heart, line segment AB is identification object plane (plane perpendicular to ground is shown as line segment in a top view)；A points and B points are respectively to know
Other thing left hand edge and right hand edge (line segment perpendicular to ground is shown as a little in a top view)；
The angle information of step 4, acquisition identification thing in binocular camera image, obtains in binocular camera image and recognizes
The angle of thing in the horizontal direction, including the horizontal sextant angle ∠ between thing left hand edge and right hand edge is recognized in left camera image
AOB is set to θ1, identification thing right hand edge and the positive direction horizontal sextant angle ∠ BOO of x-axis in left camera image1It is set to θ2, right camera
The horizontal sextant angle ∠ AO between thing left hand edge and right hand edge are recognized in image1B is set to θ3, the identification thing left side in right camera image
Edge and the horizontal sextant angle ∠ AO of x-axis negative direction1O is set to θ4；
Step 5, the distance for calculating identification thing edge and camera photocentre, with left camera photocentre O, right camera photocentre O1
The triangle Δ OO surrounded in the horizontal direction with identification thing left hand edge A1A, is obtained according to triangle sine,
In formula, l1Represent right camera photocentre O1Horizontal range with recognizing thing left hand edge A, i.e. line segment AO1Length, d tables
Show the distance of camera photocentre in left and right in binocular camera, i.e. line segment OO1Length, right camera photocentre O is solved by formula (1)1
Horizontal range l with recognizing thing left hand edge A1For,
Similarly, with left camera photocentre O, right camera photocentre O1Surrounded in the horizontal direction with identification thing right hand edge B
Triangle Δ OO1B, is obtained according to triangle sine,
In formula, l2Represent right camera photocentre O1Horizontal range with recognizing thing right hand edge B, i.e. line segment BO1Length, by
Formula (3) solves right camera photocentre O1Horizontal range l with recognizing thing right hand edge B2For,
Step 6, calculating recognize level angle information of the thing with respect to binocular camera, with right camera photocentre O1, identification thing
Left hand edge A is with recognizing thing right hand edge B surrounded triangle Δ AO in the horizontal direction1B, is obtained according to the triangle cosine law,
l2=l1 2+l2 2-2l1·l2·cosθ3 (5)
In formula, l represents to recognize the length of thing left hand edge A and right hand edge B horizontal range, i.e. line segment AB, is solved by formula (5)
The horizontal range l that thing left hand edge A and right hand edge B must be recognized is,
Further according to triangle Δ AO1Sine is obtained in B,
In formula, θ5Represent line segment AB and line segment BO1The horizontal sextant angle ∠ ABO formed1, solved by formula (7),
Further according to triangle geometrical relationship, try to achieve identification thing is with respect to the level angle θ of x-axis,
The invention has the advantages that：A kind of horizontal relative pose recognition methods simple possible of binocular camera, take full advantage of
The characteristics of binocular vision and advantage, horizontal attitude corner just can be obtained by carrying out geometrical relationship computing to image difference.
Claims (1)
1. a kind of horizontal relative pose recognition methods of binocular camera, it is characterised in that comprise the following steps：
Step 1, installation identification thing and binocular camera, fix an identification on the object for needing the horizontal relative pose of identification
Thing, identification thing needs the condition met to be to have two linear edges perpendicular to the ground, by binocular camera horizontal positioned, then
Identification thing is placed in the image pickup scope of binocular camera, to ensure the image information that can collect identification thing；
The binocular image of step 2, collection identification thing, opens operation camera, synchronization pair is gathered out by binocular camera
Recognize the image of two different azimuths of thing；
Step 3, structure plane coordinate system, the straight line where crossing the left and right camera photocentre of binocular camera are built parallel to ground
Plane coordinate system (x, y), using the straight line where the camera photocentre line of left and right as x-axis, left camera photocentre points to right shooting
Head photocentre is x-axis positive direction；Using excessively left camera photocentre and perpendicular to x-axis horizontal direction as y-axis, point to identification thing side
Direction is y-axis positive direction；Here, also it is left camera photocentre, O if O points are coordinate origin1Point is right camera photocentre, line
Section AB is identification object plane, and the plane perpendicular to ground is shown as line segment in a top view；A points and B points are respectively that identification thing is left
Edge and right hand edge, the line segment perpendicular to ground are shown as a little in a top view；
The angle information of step 4, acquisition identification thing in binocular camera image, obtains identification thing in binocular camera image and exists
Recognize that the horizontal sextant angle ∠ AOB between thing left hand edge and right hand edge are set in angle in horizontal direction, including left camera image
For θ1, identification thing right hand edge and the positive direction horizontal sextant angle ∠ BOO of x-axis in left camera image1It is set to θ2, right camera image
Horizontal sextant angle ∠ AO between middle identification thing left hand edge and right hand edge1B is set to θ3, identification thing left hand edge and x in right camera image
The horizontal sextant angle ∠ AO of axle negative direction1O is set to θ4；
Step 5, the distance for calculating identification thing edge and camera photocentre, with left camera photocentre O, right camera photocentre O1With knowledge
The triangle Δ OO that other thing left hand edge A is surrounded in the horizontal direction1A, is obtained according to triangle sine,
In formula, l1Represent right camera photocentre O1Horizontal range with recognizing thing left hand edge A, i.e. line segment AO1Length, d represents double
The distance of left and right camera photocentre, i.e. line segment OO in mesh camera1Length, right camera photocentre O is solved by formula (1)1With knowledge
Other thing left hand edge A horizontal range l1For,
Similarly, with left camera photocentre O, right camera photocentre O1The triangle surrounded in the horizontal direction with identification thing right hand edge B
Shape Δ OO1B, is obtained according to triangle sine,
In formula, l2Represent right camera photocentre O1Horizontal range with recognizing thing right hand edge B, i.e. line segment BO1Length, by formula
(3) right camera photocentre O is solved1Horizontal range l with recognizing thing right hand edge B2For,
Step 6, calculating recognize level angle information of the thing with respect to binocular camera, with right camera photocentre O1, identification thing left hand edge
A is with recognizing thing right hand edge B surrounded triangle Δ AO in the horizontal direction1B, is obtained according to the triangle cosine law,
l2=l1 2+l2 2-2l1·l2·cosθ3 (5)
In formula, l is represented to recognize the length of thing left hand edge A and right hand edge B horizontal range, i.e. line segment AB, and knowledge is solved by formula (5)
Other thing left hand edge A and right hand edge B horizontal range l is,
Further according to triangle Δ AO1Sine is obtained in B,
In formula, θ5Represent line segment AB and line segment BO1The horizontal sextant angle ∠ ABO formed1, solved by formula (7),
Further according to triangle geometrical relationship, try to achieve identification thing is with respect to the level angle θ of x-axis,
。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710219009.5A CN107121131B (en) | 2017-04-06 | 2017-04-06 | A kind of horizontal relative pose recognition methods of binocular camera |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710219009.5A CN107121131B (en) | 2017-04-06 | 2017-04-06 | A kind of horizontal relative pose recognition methods of binocular camera |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107121131A true CN107121131A (en) | 2017-09-01 |
| CN107121131B CN107121131B (en) | 2019-06-25 |
Family
ID=59726152
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710219009.5A Active CN107121131B (en) | 2017-04-06 | 2017-04-06 | A kind of horizontal relative pose recognition methods of binocular camera |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107121131B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110477956A (en) * | 2019-09-27 | 2019-11-22 | 哈尔滨工业大学 | A kind of intelligent checking method of the robotic diagnostic system based on ultrasound image guidance |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090297020A1 (en) * | 2008-05-29 | 2009-12-03 | Beardsley Paul A | Method and system for determining poses of semi-specular objects |
| US20100166294A1 (en) * | 2008-12-29 | 2010-07-01 | Cognex Corporation | System and method for three-dimensional alignment of objects using machine vision |
| CN102506757A (en) * | 2011-10-10 | 2012-06-20 | 南京航空航天大学 | Self-positioning method of binocular stereo measuring system in multiple-visual angle measurement |
| CN103512558A (en) * | 2013-10-08 | 2014-01-15 | 北京理工大学 | Binocular video pose measuring method for conical target and target patterns |
| CN105910535A (en) * | 2016-05-10 | 2016-08-31 | 天津大学 | Monocular and binocular pose deviation measuring method used for cabin automatic docking |
-
2017
- 2017-04-06 CN CN201710219009.5A patent/CN107121131B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090297020A1 (en) * | 2008-05-29 | 2009-12-03 | Beardsley Paul A | Method and system for determining poses of semi-specular objects |
| US20100166294A1 (en) * | 2008-12-29 | 2010-07-01 | Cognex Corporation | System and method for three-dimensional alignment of objects using machine vision |
| CN102506757A (en) * | 2011-10-10 | 2012-06-20 | 南京航空航天大学 | Self-positioning method of binocular stereo measuring system in multiple-visual angle measurement |
| CN103512558A (en) * | 2013-10-08 | 2014-01-15 | 北京理工大学 | Binocular video pose measuring method for conical target and target patterns |
| CN105910535A (en) * | 2016-05-10 | 2016-08-31 | 天津大学 | Monocular and binocular pose deviation measuring method used for cabin automatic docking |
Non-Patent Citations (1)
| Title |
|---|
| 于永军，徐锦法，张梁，熊智: "惯导／双目视觉位姿估计算法研究", 《仪器仪表学报》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110477956A (en) * | 2019-09-27 | 2019-11-22 | 哈尔滨工业大学 | A kind of intelligent checking method of the robotic diagnostic system based on ultrasound image guidance |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107121131B (en) | 2019-06-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109308693B (en) | Single-binocular vision system for target detection and pose measurement constructed by one PTZ camera | |
| US10899014B2 (en) | Multiple lens-based smart mechanical arm and positioning and assembly method thereof | |
| WO2019105044A1 (en) | Method and system for lens distortion correction and feature extraction | |
| CN104463108B (en) | A kind of monocular real time target recognitio and pose measuring method | |
| CN111121655B (en) | Visual detection method for pose and aperture of coplanar workpiece with equal large hole patterns | |
| CN102589530B (en) | Method for measuring position and gesture of non-cooperative target based on fusion of two dimension camera and three dimension camera | |
| CN103278090A (en) | Visual measurement method for volume of irregular object | |
| CN104034269B (en) | A kind of monocular vision measuring method and device | |
| CN103292695A (en) | Monocular stereoscopic vision measuring method | |
| CN111127540B (en) | Automatic distance measurement method and system for three-dimensional virtual space | |
| CN104700385B (en) | The binocular visual positioning device realized based on FPGA | |
| CN106991705B (en) | Position parameter estimation method based on P3P algorithm | |
| CN107084680A (en) | A kind of target depth measuring method based on machine monocular vision | |
| CN103106661A (en) | Solving parabolic catadioptric camera parameters through two intersected straight lines in space | |
| CN107816942A (en) | A kind of planar dimension measurement method based on cross structure light vision system | |
| CN109636790B (en) | Pipeline structure identification method and device | |
| CN106504287A (en) | Monocular vision object space alignment system based on template | |
| CN107957246A (en) | Article geometrical size measuring method on conveyer belt based on binocular vision | |
| CN102930548A (en) | Method for linearly solving camera inner parameter by using two identical intersected ellipses | |
| CN112362034B (en) | Solid engine multi-cylinder section butt joint guiding measurement method based on binocular vision | |
| CN106651957B (en) | Monocular vision object space localization method based on template | |
| CN103116892A (en) | Linear solving of video camera internal parameters by utilizing two intersected same circles and common tangent | |
| CN104515502A (en) | Robot hand-eye stereo vision measurement method | |
| CN110954555A (en) | WDT 3D vision detection system | |
| CN103176606A (en) | Plane interaction system and method based on binocular vision recognition |
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 |