CN109855602A - Move the monocular visual positioning method under visual field - Google Patents
Move the monocular visual positioning method under visual field Download PDFInfo
- Publication number
- CN109855602A CN109855602A CN201910030208.0A CN201910030208A CN109855602A CN 109855602 A CN109855602 A CN 109855602A CN 201910030208 A CN201910030208 A CN 201910030208A CN 109855602 A CN109855602 A CN 109855602A
- Authority
- CN
- China
- Prior art keywords
- road sign
- coordinate system
- camera
- coordinate
- video camera
- 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.)
- Pending
Links
Abstract
The invention discloses a kind of monocular visual positioning methods in the case where moving visual field, camera calibration is carried out to video camera first, it obtains camera intrinsic parameter and distortion parameter and distortion correction is carried out to image, then it is detected using video camera and identifies the vision road sign designed, to obtain road sign position information, by introducing road sign coordinate system in three-dimensional localization model, position of the video camera relative to road sign is determined by coordinate transform, finally the coordinate using road sign under camera coordinate system and world coordinates form two orientational vectors, the rotation and translation relationship of two coordinate systems is obtained by orientational vector vector product, to obtain the world coordinates of video camera, that is the position of mobile robot.The present invention compensates for the deficiency of the range information missing of monocular vision, realizes under the movement visual field environment formed in video camera with moveable robot movement, only relies on a video camera progress Mobile robot self-localization.
Description
Technical field
The present invention relates to a kind of monocular visual positioning methods in the case where moving visual field, belong to digital image processing techniques neck
Domain.
Background technique
Since machine vision technique has many advantages, such as to contain much information, flexibility is high, at low cost, vision positioning technology has been at present
Key technology and research hotspot as autonomous mobile robot.It is close several since monocular-camera is for the missing of range information
Year, the monocular vision location technology of view-based access control model road sign was made slow progress, and conventional method is fixed video camera, is formed changeless
Visual field environment obtains camera inside and outside parameter by camera calibration, to be coordinately transformed realization target machine by imaging model
The positioning of device people.In the case where video camera is with moveable robot movement, the visual field environment of movement is formd, in such case
Lower traditional monocular vision location technology is since range information missing is restricted, although binocular visual positioning technology and RGB-D
The use of video camera compensates for the missing of range information, but causes location algorithm complexity, inefficiency, and distance of taking pictures is limited
The defects of.
Summary of the invention
The purpose of the present invention is to provide a kind of monocular visual positioning methods in the case where moving visual field, have been mainly concerned with number
Word image processing techniques.
The technical solution of the invention is as follows:
A kind of monocular visual positioning method in the case where moving visual field, it is characterized in that: camera calibration is carried out to video camera first,
It obtains camera intrinsic parameter and distortion parameter and distortion correction is carried out to image, then detected and identified using video camera and designed
Vision road sign by introducing road sign coordinate system in three-dimensional localization model, become by coordinate to obtain road sign position information
Position of the determining video camera relative to road sign is changed, finally the coordinate using road sign under camera coordinate system and world coordinates are formed
Two orientational vectors obtain the rotation and translation relationship of two coordinate systems by orientational vector vector product, to obtain video camera
World coordinates, i.e. the position of mobile robot.
Specific method includes the following steps:
Step 1) carries out camera calibration to video camera first, obtains the inner parameter matrix and distortion parameter of camera；
Step 2) carries out distortion correction to the vision road sign image taken using the distortion parameter in step 1), then sharp
Road sign is detected with the contour structure of road sign, to extract road sign image；
For step 3) by Hough transform algorithm and inverse perspective mapping algorithm, logo image of satisfying the need carries out rotation and Slant Rectify,
Then road sign is identified using the arc angle information inside road sign, obtain the position letter of road sign central point and angle point in world coordinate system
Breath；
Road sign is calculated in three-dimensional localization model, through the information that coordinate transform and identification road sign obtain in step 4)
The coordinate of central point and angle point under camera coordinate system；The Y-axis of camera coordinate system is parallel with world coordinate system Y-axis；
The world that step 5) is obtained using the coordinate of central point and any angle point under camera coordinate system with identification road sign
Coordinate is respectively formed an orientational vector under two coordinate systems, and video camera is calculated by two orientational vector vector products and sits
The rotation relationship of mark system and world coordinate system；
Step 6) introduces road sign coordinate system in three-dimensional localization model；There is only flat for road sign coordinate system and camera coordinate system
Rotation relationship is not present in shifting relationship, and Y-axis is parallel with world coordinate system Y-axis, therefore is easy to get video camera under road sign coordinate system
Coordinate and road sign coordinate system and world coordinate system rotation and translation relationship；Camera shooting can be obtained finally by matrixing
The world coordinates of machine, the i.e. world coordinates of mobile robot.
Camera horizon is mounted in mobile robot, and can be rotated horizontally.
In three-dimensional localization model, road sign central point is calculated by the information that coordinate transform and identification road sign obtain
With coordinate of the angle point under camera coordinate system.
Road sign is detected using road sign contour structure, to extract road sign image and detect the figure of road sign central point and angle point
As coordinate, then identify that road sign obtains the location information of road sign central point and angle point in world coordinate system.
The present invention carries out accurate detection and identification by combining Digital image processing technique, to designed road sign, then
The deficiency that the range information missing of monocular vision is compensated for by the location information of road sign and the road sign coordinate system of introducing, realizes
Under the movement visual field environment formed in video camera with moveable robot movement, only relies on a video camera and carry out mobile robot
It is self-positioning, the autonomous fixed of the mobile robot of horizontal assembly video camera can be widely applied to based on method designed by the present invention
Position and navigation system.
Detailed description of the invention
Present invention will be further explained below with reference to the attached drawings and examples.
Fig. 1 is the vision road sign of the designed monocular visual positioning method in the case where moving visual field of the present invention；
Fig. 2 is the block schematic illustration of the designed monocular visual positioning method in the case where moving visual field of the present invention；
Fig. 3 is the camera imaging model figure of the designed monocular visual positioning method in the case where moving visual field of the present invention；
Fig. 4 is the three-dimensional localization illustraton of model of the designed monocular visual positioning method in the case where moving visual field of the present invention.
Specific embodiment
Specific embodiments of the present invention will be described in further detail with reference to the accompanying drawings of the specification.
In view of the range information of monocular vision location technology lacks problem, camera horizon is installed in the present invention and is guaranteed
Camera coordinate system is parallel with the Y-axis of world coordinate system, then solves this using the location information that identification road sign obtains and asks
Topic, so that coordinate of the road sign central point under camera coordinate system can be obtained using coordinate transform in three-dimensional localization model；This
Invention also passes through the coordinate of road sign central point and any angle point under camera coordinate system and identifies the world coordinates that road sign obtains
An orientational vector is respectively formed under two coordinate systems, to acquire two coordinate systems using two orientational vector vector products
Rotation relationship, and then the rotation and translation matrix of road sign coordinate system and world coordinate system can be obtained, to realize that monocular vision is fixed
Position.
As shown in Fig. 2, the present invention relates to a kind of monocular visual positioning methods in the case where moving visual field, in video camera with shifting
Mobile robot moves under the movement visual field environment to be formed, and road sign is detected and extracted using road sign contour structure, by identifying road sign
Location information of the road sign under world coordinate system is obtained, road sign coordinate system is introduced in three-dimensional localization model, makes road sign coordinate system
Y-axis it is parallel with the Y-axis of world coordinate system, to obtain coordinate of the road sign under camera coordinate system using coordinate transform, then
The rotation and translation matrix of road sign coordinate system and world coordinate system is obtained using orientational vector vector product, it is final by matrixing
Obtain coordinate of the video camera under world coordinate system, the i.e. location information of mobile robot indoors.In practical applications, it uses
Robot camera shoots vision road sign image, and camera horizon is mounted in mobile robot, and can be rotated horizontally,
Use Matlab, OpenCV and VS2013 software realization monocular vision location algorithm.
As shown in figure 3, executing following steps:
Step 1) carries out camera calibration to video camera first, obtains the inner parameter matrix and distortion parameter of camera；
Specific demarcating steps are as follows:
As shown in figure 3, being related to 4 kinds of coordinate systems in the imaging model of video camera, respectively imaging coordinate system, image are sat
Mark system, camera coordinate system and world coordinate system, obtain image coordinate system and world coordinates by the conversion of each coordinate system
Shown in mapping relations such as formula (1) between system:
Wherein, f indicates that focal length of camera, dx, dy respectively indicate the physical size of each pixel on x-axis and y-axis direction,
(u0,v0) it is camera optical center OcCoordinate under image coordinate system, M1For the resulting inner parameter matrix of camera calibration, M2For phase
Machine demarcates resulting external parameter.
Shown in the model of the radial distortion of video camera such as formula (2):
Shown in the model of the tangential error of video camera such as formula (3):
Zhang Zhengyou camera calibration method is used based on the above principles, does calibrating template with 4 × 6 gridiron pattern, is finally obtained interior
Parameter matrix M1With distortion parameter k1,k2,k3,p1,p2。
Step 2) carries out distortion correction to the vision road sign image taken using the distortion parameter in step 1), then sharp
Road sign is detected with the contour structure of road sign, to extract road sign image；
It is as shown in Figure 1 designed vision road sign, the distance ratio of road sign nested inside outer profile is 1:2:2, passes through wheel
Exterior feature detection and distance detect than limiting and extract road sign image.
For step 3) by Hough transform algorithm and inverse perspective mapping algorithm, logo image of satisfying the need carries out rotation and Slant Rectify,
Then road sign is identified using the arc angle information inside road sign, obtain the position letter of road sign central point and angle point in world coordinate system
Breath；
Specific identification step is as follows:
First using four vertex of Hough transform algorithm detection road sign dark border, then become using based on corresponding points
The inverse perspective mapping changed satisfy the need logo image carry out rotation and Slant Rectify.
Then it is pre-processed using Gaussian filter algorithm and the algorithm for image enhancement based on fuzzy field logo image of satisfying the need, benefit
Road sign is split with contour detecting and Least Square Circle fitting process.
Finally the road sign image after segmentation is identified and decoded using arc angle information, to obtain the letter of road sign carrying
Breath, i.e., then at the location information of world coordinate system road sign angle can be obtained using road sign actual physical size in road sign central point
The location information of point.
Road sign is calculated in three-dimensional localization model, through the information that coordinate transform and identification road sign obtain in step 4)
The coordinate of central point and angle point under camera coordinate system (Y-axis is parallel with world coordinate system Y-axis)；
Specific step is as follows:
As shown in figure 4, in three-dimensional localization model, since camera coordinate system Y-axis is parallel with world coordinate system Y-axis, institute
The y of road sign coordinate under camera coordinate system can be determined with the world coordinates by road signcValue, it is then public using coordinate transform
(x can be obtained in formula (4)c/zc,yc/zc, 1), coordinate (x of the road sign under camera coordinate system can be obtainedc,yc,zc)。
Similarly, coordinate of the road sign angle point under camera coordinate system can be obtained using above-mentioned steps.
The world that step 5) is obtained using the coordinate of central point and any angle point under camera coordinate system with identification road sign
Coordinate is respectively formed an orientational vector under two coordinate systems, and camera coordinate system is calculated by orientational vector vector product
With the rotation relationship of world coordinate system；
Specific step is as follows:
Coordinate first with road sign central point A and angle point B under camera coordinate system forms a vectorUse road sign
The world coordinates of central point A and angle point B form another vectorThen it is calculated and takes the photograph by formula (5) and formula (6)
The rotation angle of camera coordinate system and world coordinate systemCamera coordinate system and world coordinate system are obtained by formula (7)
Spin matrix R (A → B).
Step 6) introduces road sign coordinate system in three-dimensional localization model, and (there is only flat for road sign coordinate system and camera coordinate system
Shifting relationship, is not present rotation relationship, and Y-axis is parallel with world coordinate system Y-axis), therefore video camera is easy to get in road sign coordinate system
Under coordinate and road sign coordinate system and world coordinate system rotation and translation relationship, can be obtained and take the photograph finally by matrixing
The world coordinates of camera, the i.e. world coordinates of mobile robot.
The monocular visual positioning method in the case where moving visual field that the present invention designs, by combining Digital image processing technique,
Accurate detection and identification are carried out to designed road sign, then made up by the location information of road sign and the road sign coordinate system of introducing
The deficiency of the range information missing of monocular vision, realizes the movement visual field formed in video camera with moveable robot movement
Under environment, only relies on a video camera and carry out Mobile robot self-localization, can be answered extensively based on method designed by the present invention
For being horizontally mounted the autonomous positioning and navigation system of the mobile robot of single camera.
Embodiments of the present invention are explained in detail above in conjunction with attached drawing, but the present invention is not limited to above-mentioned implementations
Mode within the knowledge of a person skilled in the art can also be without departing from the purpose of the present invention
It makes a variety of changes.
Claims (5)
1. a kind of monocular visual positioning method in the case where moving visual field obtains it is characterized in that: carrying out camera calibration to video camera first
Distortion correction is carried out to camera intrinsic parameter and distortion parameter and to image, is then detected and is identified using video camera and designed
Vision road sign, by introducing road sign coordinate system in three-dimensional localization model, passes through coordinate transform to obtain road sign position information
Determine position of the video camera relative to road sign, finally the coordinate using road sign under camera coordinate system and world coordinates form two
A orientational vector obtains the rotation and translation relationship of two coordinate systems by orientational vector vector product, to obtain video camera
World coordinates, the i.e. position of mobile robot.
2. the monocular visual positioning method according to claim 1 in the case where moving visual field, it is characterized in that: specific method includes
The following steps:
Step 1) carries out camera calibration to video camera first, obtains the inner parameter matrix and distortion parameter of camera；
Step 2 carries out distortion correction to the vision road sign image taken using the distortion parameter in step 1), then utilizes road
Target contour structure detects road sign, to extract road sign image；
Step 3) is by Hough transform algorithm and inverse perspective mapping algorithm, and logo image of satisfying the need carries out rotation and Slant Rectify, then
Road sign is identified using the arc angle information inside road sign, obtains the location information of road sign central point and angle point in world coordinate system；
Road sign center is calculated in three-dimensional localization model, through the information that coordinate transform and identification road sign obtain in step 4)
Point and coordinate of the angle point under camera coordinate system；The Y-axis of camera coordinate system is parallel with world coordinate system Y-axis；
The world coordinates that step 5) is obtained using the coordinate of central point and any angle point under camera coordinate system with identification road sign
It is respectively formed an orientational vector under two coordinate systems, camera coordinate system is calculated by two orientational vector vector products
With the rotation relationship of world coordinate system；
Step 6) introduces road sign coordinate system in three-dimensional localization model；There is only translations to close for road sign coordinate system and camera coordinate system
Rotation relationship is not present in system, and Y-axis is parallel with world coordinate system Y-axis, therefore is easy to get seat of the video camera under road sign coordinate system
It is marked with and the rotation and translation relationship of road sign coordinate system and world coordinate system；Video camera can be obtained finally by matrixing
World coordinates, the i.e. world coordinates of mobile robot.
3. the monocular visual positioning method according to claim 2 in the case where moving visual field, it is characterized in that: camera horizon is pacified
In mobile robot, and it can be rotated horizontally.
4. the monocular visual positioning method according to claim 2 in the case where moving visual field, it is characterized in that: in three-dimensional localization mould
In type, road sign central point and angle point are calculated in camera coordinate system by the information that coordinate transform and identification road sign obtain
Under coordinate.
5. the monocular visual positioning method according to claim 2 in the case where moving visual field, it is characterized in that: utilizing road sign profile
Then structure detection road sign identifies road sign to extract road sign image and detect the image coordinate of road sign central point and angle point
Obtain the location information of road sign central point and angle point in world coordinate system.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910030208.0A CN109855602A (en) | 2019-01-14 | 2019-01-14 | Move the monocular visual positioning method under visual field |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910030208.0A CN109855602A (en) | 2019-01-14 | 2019-01-14 | Move the monocular visual positioning method under visual field |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109855602A true CN109855602A (en) | 2019-06-07 |
Family
ID=66894605
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910030208.0A Pending CN109855602A (en) | 2019-01-14 | 2019-01-14 | Move the monocular visual positioning method under visual field |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109855602A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110619664A (en) * | 2019-09-17 | 2019-12-27 | 武汉理工大学 | Camera distance and attitude calculation method based on laser pattern assistance and server |
| CN110910410A (en) * | 2019-11-07 | 2020-03-24 | 河海大学 | Court positioning system and method based on computer vision |
| CN111536984A (en) * | 2020-05-11 | 2020-08-14 | 北京百度网讯科技有限公司 | Positioning method and device, vehicle-end equipment, vehicle, electronic equipment and positioning system |
| CN111968177A (en) * | 2020-07-22 | 2020-11-20 | 东南大学 | Mobile robot positioning method based on fixed camera vision |
| CN113268063A (en) * | 2021-06-03 | 2021-08-17 | 北京京东乾石科技有限公司 | Control method and device for robot and non-volatile computer readable storage medium |
| CN113345024A (en) * | 2020-02-18 | 2021-09-03 | 致伸科技股份有限公司 | Method for judging assembling quality of camera module |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2372651A1 (en) * | 2010-03-08 | 2011-10-05 | Ricoh Company, Ltd. | Image pickup apparatus and range determination system |
| CN102419178A (en) * | 2011-09-05 | 2012-04-18 | 中国科学院自动化研究所 | Mobile robot positioning system and method based on infrared road sign |
| CN106092090A (en) * | 2016-08-06 | 2016-11-09 | 中科院合肥技术创新工程院 | A kind of infrared road sign for indoor mobile robot location and using method thereof |
| CN106969766A (en) * | 2017-03-21 | 2017-07-21 | 北京品创智能科技有限公司 | A kind of indoor autonomous navigation method based on monocular vision and Quick Response Code road sign |
| CN107832661A (en) * | 2017-09-27 | 2018-03-23 | 南通大学 | A kind of Localization Approach for Indoor Mobile of view-based access control model road sign |
| CN108571971A (en) * | 2018-05-17 | 2018-09-25 | 北京航空航天大学 | A kind of AGV vision positioning systems and method |
| CN108759773A (en) * | 2018-05-07 | 2018-11-06 | 深圳市三宝创新智能有限公司 | A kind of monocular vision distance measuring method applied to robot crawl |
-
2019
- 2019-01-14 CN CN201910030208.0A patent/CN109855602A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2372651A1 (en) * | 2010-03-08 | 2011-10-05 | Ricoh Company, Ltd. | Image pickup apparatus and range determination system |
| CN102419178A (en) * | 2011-09-05 | 2012-04-18 | 中国科学院自动化研究所 | Mobile robot positioning system and method based on infrared road sign |
| CN106092090A (en) * | 2016-08-06 | 2016-11-09 | 中科院合肥技术创新工程院 | A kind of infrared road sign for indoor mobile robot location and using method thereof |
| CN106969766A (en) * | 2017-03-21 | 2017-07-21 | 北京品创智能科技有限公司 | A kind of indoor autonomous navigation method based on monocular vision and Quick Response Code road sign |
| CN107832661A (en) * | 2017-09-27 | 2018-03-23 | 南通大学 | A kind of Localization Approach for Indoor Mobile of view-based access control model road sign |
| CN108759773A (en) * | 2018-05-07 | 2018-11-06 | 深圳市三宝创新智能有限公司 | A kind of monocular vision distance measuring method applied to robot crawl |
| CN108571971A (en) * | 2018-05-17 | 2018-09-25 | 北京航空航天大学 | A kind of AGV vision positioning systems and method |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110619664A (en) * | 2019-09-17 | 2019-12-27 | 武汉理工大学 | Camera distance and attitude calculation method based on laser pattern assistance and server |
| CN110619664B (en) * | 2019-09-17 | 2023-06-27 | 武汉理工大学 | Laser pattern-assisted camera distance posture calculation method and server |
| CN110910410A (en) * | 2019-11-07 | 2020-03-24 | 河海大学 | Court positioning system and method based on computer vision |
| CN113345024A (en) * | 2020-02-18 | 2021-09-03 | 致伸科技股份有限公司 | Method for judging assembling quality of camera module |
| CN113345024B (en) * | 2020-02-18 | 2024-03-05 | 致伸科技股份有限公司 | Method for judging assembly quality of camera module |
| CN111536984A (en) * | 2020-05-11 | 2020-08-14 | 北京百度网讯科技有限公司 | Positioning method and device, vehicle-end equipment, vehicle, electronic equipment and positioning system |
| US11405744B2 (en) | 2020-05-11 | 2022-08-02 | Apollo Intelligent Connectivity (Beijing) Technology Co., Ltd. | Positioning method and device, on-board equipment, vehicle, and positioning system |
| CN111968177A (en) * | 2020-07-22 | 2020-11-20 | 东南大学 | Mobile robot positioning method based on fixed camera vision |
| CN113268063A (en) * | 2021-06-03 | 2021-08-17 | 北京京东乾石科技有限公司 | Control method and device for robot and non-volatile computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109855602A (en) | Move the monocular visual positioning method under visual field | |
| CN105758426B (en) | The combined calibrating method of the multisensor of mobile robot | |
| CN111750820B (en) | Image positioning method and system | |
| CN107274336B (en) | A kind of Panorama Mosaic method for vehicle environment | |
| CN106228538B (en) | Binocular vision indoor orientation method based on logo | |
| CN109035200A (en) | A kind of bolt positioning and position and posture detection method based on the collaboration of single binocular vision | |
| CN110842940A (en) | Building surveying robot multi-sensor fusion three-dimensional modeling method and system | |
| CN108460779A (en) | A kind of mobile robot image vision localization method under dynamic environment | |
| JP5872818B2 (en) | Positioning processing device, positioning processing method, and image processing device | |
| CN100501327C (en) | Single-vision measuring method of space three-dimensional attitude of variable-focus video camera | |
| CN108827316A (en) | Mobile robot visual orientation method based on improved Apriltag label | |
| CN106683137B (en) | Artificial mark based monocular and multiobjective identification and positioning method | |
| CN109767476A (en) | A kind of calibration of auto-focusing binocular camera and depth computing method | |
| CN109859272A (en) | A kind of auto-focusing binocular camera scaling method and device | |
| CN108305264A (en) | A kind of unmanned plane precision landing method based on image procossing | |
| CN101763643A (en) | Automatic calibration method for structured light three-dimensional scanner system | |
| CN110174088A (en) | A kind of target ranging method based on monocular vision | |
| Carr et al. | Point-less calibration: Camera parameters from gradient-based alignment to edge images | |
| CN110009682A (en) | A kind of object recognition and detection method based on monocular vision | |
| CN109801360B (en) | Image-based gastrointestinal three-dimensional reconstruction and visualization method | |
| CN108470356A (en) | A kind of target object fast ranging method based on binocular vision | |
| CN107832661B (en) | A kind of Localization Approach for Indoor Mobile of view-based access control model road sign | |
| CN109615645A (en) | The Feature Points Extraction of view-based access control model | |
| CN112200203A (en) | Matching method of weak correlation speckle images in oblique field of view | |
| CN106952262B (en) | Ship plate machining precision analysis method based on stereoscopic vision |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190607 |
|
| RJ01 | Rejection of invention patent application after publication |