CN107885224A - Unmanned plane barrier-avoiding method based on tri-item stereo vision - Google Patents
Unmanned plane barrier-avoiding method based on tri-item stereo vision Download PDFInfo
- Publication number
- CN107885224A CN107885224A CN201711078365.6A CN201711078365A CN107885224A CN 107885224 A CN107885224 A CN 107885224A CN 201711078365 A CN201711078365 A CN 201711078365A CN 107885224 A CN107885224 A CN 107885224A
- Authority
- CN
- China
- Prior art keywords
- line
- barrier
- triple
- matching
- unmanned plane
- 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
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course or altitude of land, water, air, or space vehicles, e.g. automatic pilot
- G05D1/10—Simultaneous control of position or course in three dimensions
- G05D1/101—Simultaneous control of position or course in three dimensions specially adapted for aircraft
-
- 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
Abstract
The invention provides a kind of unmanned plane barrier-avoiding method based on tri-item stereo vision, including three mesh stereoscopic cameras of demarcation, the intrinsic parameter of three cameras of acquisition and outer parameter;Image is gathered, obtains three-view diagram;The characteristic straight line in three-view diagram is extracted, establishes matching line segments triple, the locus of line style barrier is estimated according to matching line segments triple;The angle point in three-view diagram is extracted, establishes Feature Points Matching triple, the locus of non-linearity barrier is estimated according to Feature Points Matching triple;According to the locus of line style barrier and the locus of non-linearity barrier, Obstacles distribution map is established;According to Obstacles distribution map, acquisition can traffic areas, planning unmanned plane safe flight path.Which raises the precision of detection of obstacles and the security for improving unmanned plane during flying.
Description
Technical field
The present invention relates to unmanned air vehicle technique field, more particularly to the unmanned plane barrier-avoiding method based on tri-item stereo vision.
Background technology
Detection of obstacles technology is one of key technology in unmanned plane information Perception system, fast and accurately detection flight
Barrier in environment is the basis of the safe autonomous flight of unmanned plane.The unmanned plane avoidance technology of view-based access control model has investigative range
The wide, advantage such as information capacity is big, cost is low, it is especially rapid to flight environment of vehicle change capture, reaction is sharp the features such as, make its
Increasing concern has been obtained in unmanned plane avoidance technical research.
Monocular or binocular stereo vision can be used in the detection of obstacles of view-based access control model, and widely used at present is binocular solid
Vision avoidance.It is auxiliary that Publication No. CN103411526A Chinese patent application discloses a kind of driving based on binocular stereo vision
Help obstacle detection method.This method utilizes Feature Points Matching result, and barrier is detected based on binocular reconfiguration principle.Similar,
Publication No. CN102313536A Chinese patent application discloses a kind of method for barrier perception based on airborne binocular vision.
This method carries out barrier perception using matching double points generation disparity map according to disparity map.Such method is not suitable for existing
Unmanned plane avoidance in the linear obstacle environment such as power line.
In electric power line inspection field, there are many electric power line detecting methods.For example, Publication No. CN103984355A China
A kind of inspection flying robot of patent application publication and overhead power line range prediction and keeping method.Its power line distance
Method is to obtain geographical position coordinates by airborne global positioning system and be transformed into power line threedimensional model coordinate system, is utilized
Three-dimensional corridor model estimation flying robot and the distance of power line.The major defect of this method is to rely on known power line
Three-dimensional corridor model, is not suitable for not modeling region.Publication No. CN103810462A Chinese patent application discloses one kind
Interrupt method based on linear target.Composition is approximately put in Hessian matrix characteristic vectors direction by this method
Small line segment, small line segment is being connected as long straight line using line fitting method, the position of algorithm only detection of straight lines in the picture, and
The locus of straight line is not provided.
Analyzed more than, subject matter existing for the detection of obstacles technology of existing view-based access control model is:(1) it is more
Number vision avoidance detection algorithm only relies upon Feature Points Matching to detect non-linearity barrier, and power line detection class algorithm is only visited
Survey line type barrier, lack the algorithm that two types barrier can all detect;(2) even if consider in the presence of a small number of obstacle avoidance algorithms simultaneously
Line style and non-linearity barrier, but depend on the scheme of binocular stereo vision to work as there is also the risk that the leakage of line style barrier is picked up
During the basal coplane of line style barrier and binocular stereo vision, the space of line style barrier can not be estimated according to binocular imaging principle
Position.Problem above brings great security risk to autonomous flight in unmanned plane external environment out of office.
The content of the invention
In view of the above problems, it is an object of the present invention to provide one kind to overcome above mentioned problem or solve at least in part
The certainly unmanned plane barrier-avoiding method based on tri-item stereo vision of above mentioned problem.
The present invention one is further objective is that improving the precision of detection of obstacles and improving the security of unmanned plane during flying.
The invention provides a kind of unmanned plane barrier-avoiding method based on tri-item stereo vision, including:
Demarcate three mesh stereoscopic cameras, the intrinsic parameter of three cameras of acquisition and outer parameter;
Image is gathered, obtains three-view diagram;
The characteristic straight line in three-view diagram is extracted, establishes matching line segments triple, line style is estimated according to matching line segments triple
The locus of barrier;
The angle point in three-view diagram is extracted, establishes Feature Points Matching triple, is estimated according to Feature Points Matching triple non-thread
The locus of type barrier;
According to the locus of line style barrier and the locus of non-linearity barrier, Obstacles distribution is established
Figure;
According to Obstacles distribution map, acquisition can traffic areas, planning unmanned plane safe flight path.
Alternatively, the step of establishing matching line segments triple specifically includes:
For the characteristic straight line in the three-view diagram of extraction, characteristic straight line description is established using LBD algorithms;
Matching line segments triple is established according to sub- nearest neighbouring rule is described.
Alternatively, before the locus of line style barrier is estimated according to matching line segments triple, in addition to:
According to the outer parameter of three cameras, the camera matrix under structure normalization coordinate description;
The matrix of trifocal tensor is built according to camera matrix;
Line-line-line three-view diagram geometrical constraint is determined according to the matrix of trifocal tensor;
Matching line segments triple wrong in matching line segments triple is rejected according to line-line-line three-view diagram geometrical constraint, is protected
Stay correct matching line segments triple.
Alternatively, line-line-line three-view diagram geometrical constraint is:
Wherein,For matching line segments triple,ForI-th of component (i=1,2,3),It is
Predicted value, TiFor the matrix of trifocal tensor;
Wherein,WhereinWithCamera matrix P is represented respectively2And P3I-th row;
Wherein,
It is posture of the jth camera in No. i-th camera coordinates system,It is kth camera relative to jth camera
Expression of the translation vector in No. i-th camera coordinates system;And
Matching line segments triple is that the condition of correct matching line segments triple is:
Wherein, | | | | two norms of vector are represented, × represent two vectorial cross product operations.
Alternatively, the step of establishing Feature Points Matching triple specifically includes:
To the angle point in the three-view diagram of extraction, feature point description is established using ORB algorithms;
Feature Points Matching triple is established according to nearest neighbouring rule.
Alternatively, before the step of locus of non-linearity barrier is estimated according to Feature Points Matching triple, also
Including:
According to the outer parameter of three cameras, the camera matrix under structure normalization coordinate description;
The matrix of trifocal tensor is built according to camera matrix;
A point-line-three-view diagram geometrical relationship is determined according to the matrix of trifocal tensor;
Feature Points Matching ternary wrong in Feature Points Matching triple is rejected according to a point-line-three-view diagram geometrical relationship
Group, retain correct Feature Points Matching triple.
Alternatively, a point-line-three-view diagram geometrical relationship is:
Wherein,It is first j-th of characteristic point of width viewThree components of homogeneous coordinates,It isPredicted value, TiFor the matrix of trifocal tensor,WhereinWithCamera is represented respectively
Matrix P2And P3I-th row;
Wherein, P1=[I | 0], It is
Posture of the j cameras in No. i-th camera coordinates system,Be kth camera relative to the translation vector of jth camera i-th
Expression in number camera coordinates system;
It is to pass through in the 2nd width imagePut and perpendicular to the straight line to polar curve;And
Feature Points Matching triple is that the condition of correct Feature Points Matching triple is:
Wherein, | | | | two norms of vector are represented, × represent two vectorial cross product operations.
Alternatively, according to the locus of line style barrier and the locus of non-linearity barrier, barrier sky is established
Between distribution map the step of specifically include:
Place obstacles the resolution ratio of object space distribution map;
For the locus of the line style barrier of estimation, at interval of one point between two end points of line style barrier
Resolution samples a point;
The locus of the point gathered on line style barrier and the non-linearity barrier of estimation is formed into three-dimensional point cloud;
Three-dimensional point cloud is converted into Obstacles distribution map using OctoMap algorithms.
Alternatively, according to Obstacles distribution map, acquisition can traffic areas, the step in planning unmanned plane safe flight path
Suddenly specifically include:
According to Obstacles distribution map, optimal avoidance path is solved using OMPL tool storage rooms.
The unmanned plane barrier-avoiding method based on tri-item stereo vision of the present invention, line style is detected simultaneously using three mesh stereoscopic cameras
With non-linearity barrier, the unmanned plane avoidance that can be adapted to extensively in the flight environment of vehicle of field.Also, it can effectively overcome binocular solid
In vision because imaging geometry is degenerated and the defects of missing inspection line style barrier.
Further, the unmanned plane barrier-avoiding method of the invention based on tri-item stereo vision, utilizes three-view diagram geometrical constraint
The characteristic straight line and characteristic point of error hiding are removed, effectively increases the precision of detection of obstacles.
According to the accompanying drawings will be brighter to the detailed description of the specific embodiment of the invention, those skilled in the art
Above-mentioned and other purposes, the advantages and features of the present invention.
Brief description of the drawings
Some specific embodiments of the present invention are described in detail by way of example, and not by way of limitation with reference to the accompanying drawings hereinafter.
Identical reference denotes same or similar part or part in accompanying drawing.It should be appreciated by those skilled in the art that these
What accompanying drawing was not necessarily drawn to scale.In accompanying drawing:
Fig. 1 is the schematic diagram of the unmanned plane barrier-avoiding method according to an embodiment of the invention based on tri-item stereo vision;
Fig. 2 is the location diagram for three cameras that the present invention applies;
Embodiment
As shown in figure 1, the unmanned plane barrier-avoiding method based on tri-item stereo vision of the present embodiment, including:
S102, demarcate three mesh stereoscopic cameras, the intrinsic parameter of three cameras of acquisition and outer parameter.
The intrinsic parameter of three mesh stereoscopic cameras is:
KiIt is calibration matrix, whereinWithIt is the equivalent focal length in the x and y directions of i-th of camera respectively,WithRespectively
It is the coordinate of the principal point of i-th of camera on the image plane.DiContain the radially and tangentially deformation coefficient of image.It is assuming that unchanged
Shape image point coordinates is (xμ,yμ), the image coordinate after deformation is (xd,yd), relation therebetween is:
WhereinRepresent be on the plane of delineation picture point to the distance of principal point.Regarded for the three of synchronous acquisition
Figure, first with DiRadially and tangentially distortion correction is carried out to input picture.
As shown in Fig. 2 C1, C2, C3 represent the location of three cameras, it is three-dimensional by three mesh of the coordinate system of C1 cameras
The reference frame of camera, demarcating outer parameter is:
WhereinIt is posture of the jth camera in No. i-th camera coordinates system,It is kth camera relative to jth number
Expression of the translation vector of camera in No. i-th camera coordinates system.Outer parameter in above formula meets following relation:
Present embodiment assumes that the inside and outside calibrating parameters of three mesh stereoscopic cameras are obtained by scaling board and calibration tool.Mirror
In each camera intrinsic parameter, it is known that in this embodiment, following described characteristic straight lines and characteristic point are converted into normalization
Described in coordinate system, i.e., equivalent focal length is 1.
S104, image is gathered, obtain three-view diagram.
During unmanned plane during flying, three mesh stereoscopic cameras being arranged on unmanned plane gather forward image in real time, obtain three
View.
S106, the characteristic straight line in three-view diagram is extracted, establishes matching line segments triple, estimated according to matching line segments triple
The locus of line style barrier.
In view of the line style such as power line barrier is typically opening through whole image plane in the picture, it is special in this embodiment
The extraction of sign straight line uses edge connection method (Edge Drawing)+Hough transform method (Kernel-based based on Gaussian kernel
Hough Transform), the long straight line that this method can quickly in detection image.The extracting method of characteristic straight line includes:
S302, smothing filtering is carried out respectively to three-view diagram picture;
S304, it is determined that the candidate edge point of each image and its image gradient direction;
S306, determine edge anchor point and its closure;
S308, edge features are extracted according to edge anchor point and its closure;
S310, the edge cluster of near linear distribution is chosen from edge features;
S312, to each edge cluster, the Gaussian kernel of digital simulation straight line and its parameter distribution;
S314, according to the Gaussian kernel of all parameter distributions, establish the distribution map of parameter space;
S316, the peak point of Selecting All Parameters from the distribution map of parameter space, extract straight line.
For the characteristic straight line of said extracted, characteristic straight line description is established using LBD algorithms, calculates LBD description
Parameter is arranged to:Number of bands=9, Width of band=7.
Son is described using characteristic straight line, matching line segments triple is established according to sub- nearest neighbouring rule is described.Specifically, it is above-mentioned
The characteristic straight line of acquisition describes subsetWherein niRepresent that the feature in the i-th width view is straight
Line number mesh,It is j-th of characteristic straight line description in the i-th width view.The parameter setting of son is described according to LBD,It is
The vector of one 72 dimension.It is as follows according to the calculation formula that the sub- nearest neighbouring rule of description establishes matching line segments triple:
For the j-th strip characteristic straight line in the 1st width view, its matching line triple is:Wherein
Graph line is represented, graph line parametric method is in this embodiment:axμ+byμ+ c=0, i.e.,
S108, the angle point in three-view diagram is extracted, establishes Feature Points Matching triple, estimated according to Feature Points Matching triple
The locus of non-linearity barrier.
For non-linearity barrier, the distance method of the characteristic point disturbance in judgement thing by detecting blocking surfaces,
To detect more characteristic points, using ORB Corner Detections and description algorithm in this embodiment, carried from image
Take ORB characteristic points and establish ORB description.
Using feature point description, Feature Points Matching triple is established according to nearest neighbouring rule.Specifically,
Feature point description subset in three width views is obtained by ORB algorithms
Wherein niRepresent that the feature in the i-th width view is counted out,It is j-th of feature point description in the i-th width view, each retouches
State the binary string that son is 256.It is as follows according to the calculation formula that the sub- nearest neighbouring rule of description establishes Feature Points Matching triple:
For j-th of characteristic point in the 1st width view, its matching triple is:Wherein p=(xμ,
yμ,1)TRepresent characteristic point.
S110, according to the locus of line style barrier and the locus of non-linearity barrier, establish Obstacles
Distribution map.
S112, according to Obstacles distribution map, acquisition can traffic areas, planning unmanned plane safe flight path.
The matching line segments triple that S106 is established generally comprises many wrong matchings, to improve linear detection of obstacles
Accuracy rate, before the locus of line style barrier is estimated according to matching line segments triple, in addition to:
According to the outer parameter of three cameras, the camera matrix under structure normalization coordinate description.
The matrix of trifocal tensor is built by camera matrix.The matrix form of trifocal tensor is as follows:
WhereinWithCamera matrix P is represented respectively2And P3I-th row,It is that jth camera is sat in No. i-th camera
Posture in mark system,It is table of the kth camera relative to the translation vector of jth camera in No. i-th camera coordinates system
Show.
Line-line-line three-view diagram geometrical constraint is determined according to the matrix of trifocal tensor.Specifically, matching line segments tripleIt must is fulfilled for following line-line-line three-view diagram geometrical constraint:
Wherein,For the matching line segments triple,ForI-th of component (i=1,2,3),
It isPredicted value, TiFor the matrix of trifocal tensor.
Matching line segments triple wrong in matching line segments triple is rejected according to line-line-line three-view diagram geometrical constraint, is protected
Stay correct matching line segments triple.Specifically, in this embodiment ifThen think matching line segments ternary
GroupIt is correctly to match, otherwise rejects.For the correct matching line segments triple of reservation, by straight line into
As the position of model estimation space straight line.In this embodiment, space line uses Pluecker coordinate representations:Lj=(mj,
vj), wherein mjBe j-th strip space line moment vector (its direction is the normal direction that the origin of coordinates and space line determine plane, its
Distance of a length of origin of coordinates of mould to space line), vjIt is the direction vector of space line, the two satisfaction | Vj| |=1, Vj T·mj
=0.If camera matrixThe projection equation of space line is as follows:
Assuming thatEnd points in the i-th width image isThen projection error be defined as end points to projection straight line away from
From:
The object function of optimization is:
In formula, | | | | two norms of vector are represented, represent two vectorial inner products.
The initial value of iteration is obtained by the observation reconstruct of two view straight lines, then using in LM optimized algorithms solution formula (11)
Optimization problem, you can obtain j-th strip space line Lj=(mj,vj)。
The Feature Points Matching triple that S108 is established generally comprises many wrong matchings, to improve non-linearity obstacle quality testing
The accuracy rate of survey, before the step of locus of non-linearity barrier is estimated according to Feature Points Matching triple, in addition to:
According to the outer parameter of three cameras, the camera matrix under structure normalization coordinate description.
The matrix of trifocal tensor is built according to camera matrix.The matrix form of trifocal tensor is as follows:
WhereinWithCamera matrix P is represented respectively2And P3I-th row.
A point-line-three-view diagram geometrical relationship is determined according to the matrix of trifocal tensor.A point-line-three-view diagram geometrical relationship
Calculation formula is as follows:
It is to pass through in the 2nd width imagePut and perpendicular to the straight line to polar curve.
Feature Points Matching ternary wrong in Feature Points Matching triple is rejected according to a point-line-three-view diagram geometrical relationship
Group, retain correct Feature Points Matching triple.Specifically, ifThen think Feature Points Matching ternary
GroupIt is correctly to match, otherwise rejects.
For correct Feature Points Matching triple, pass through a position for imaging model estimation space point.It is real in this implementation
In example, the coordinate of spatial point is:Xj=(xj,yj,zj,1)T.If camera matrixThe projection of spatial point
Equation is as follows:
Optimization object function is defined as:
WhereinThe image observation value of characteristic point is represented, andRepresent that the image of characteristic point is pre-
Measured value, it is the function of space point coordinates, as shown in formula (14).
The initial value of iteration is obtained by the observation reconstruct of 2 view feature points, then using in LM optimized algorithms solution formula (15)
Optimization problem, you can obtain j-th of spatial point Xj=(xj,yj,zj,1)T。
The locus of line style barrier and the non-linearity barrier represented with point is obtained according to said process, for reality
Show unmanned plane avoidance, it is necessary to which discrete space line and spatial point to be converted to the Obstacles distribution map in region, and according to
Obstacles distribution map planning avoidance path.Comprise the following steps that:
Place obstacles the resolution ratio of object space distribution map.In this embodiment, the spatial distribution map OctoMap of barrier
Carry out organization and management.OctoMap is the space representation method based on Octree data organizational form, can be parameterized with probability
Can passage space, Obstacles and unknown space, have the advantages that small EMS memory occupation, map building and retrieval it is fast.Space point
Resolution can be set according to the size of unmanned plane, and in this embodiment, the resolution ratio of Obstacles distribution map is set to 1.28 meters.
For the locus of the line style barrier of estimation, at interval of one point between two end points of line style barrier
Resolution samples a point, in the present embodiment, for space line, between the two endpoints every a resolution granularity (i.e.
1.28 meters) one point of sampling.The locus of the point gathered on line style barrier and the non-linearity barrier of estimation is formed
Three-dimensional point cloud, three-dimensional point cloud is converted into Obstacles distribution map using OctoMap algorithms.
After obtaining the Obstacles distribution map that OctoMap is represented, using OMPL (Open Motion Planning
Library) motion planning storehouse solves optimal avoidance path.
The unmanned plane barrier-avoiding method based on tri-item stereo vision of the present embodiment, utilize three mesh stereoscopic cameras while detection line
Type and non-linearity barrier, the unmanned plane avoidance that can be adapted to extensively in the flight environment of vehicle of field.Also, effectively binocular can be overcome to stand
In body vision because imaging geometry is degenerated and the defects of missing inspection line style barrier.
Further, the unmanned plane barrier-avoiding method based on tri-item stereo vision of the present embodiment, using three-view diagram geometry about
Beam removes the characteristic straight line and characteristic point of error hiding, effectively increases the precision of detection of obstacles.
So far, although those skilled in the art will appreciate that detailed herein have shown and described multiple showing for the present invention
Example property embodiment, still, still can be direct according to present disclosure without departing from the spirit and scope of the present invention
It is determined that or derive many other variations or modifications for meeting the principle of the invention.Therefore, the scope of the present invention is understood that and recognized
It is set to and covers other all these variations or modifications.
Claims (9)
1. a kind of unmanned plane barrier-avoiding method based on tri-item stereo vision, including:
Demarcate three mesh stereoscopic cameras, the intrinsic parameter of three cameras of acquisition and outer parameter;
Image is gathered, obtains three-view diagram;
The characteristic straight line in the three-view diagram is extracted, establishes matching line segments triple, is estimated according to the matching line segments triple
The locus of line style barrier;
The angle point in the three-view diagram is extracted, establishes Feature Points Matching triple, is estimated according to the Feature Points Matching triple
The locus of non-linearity barrier;
According to the locus of the line style barrier and the locus of the non-linearity barrier, Obstacles point are established
Butut;
According to the Obstacles distribution map, acquisition can traffic areas, planning unmanned plane safe flight path.
2. unmanned plane barrier-avoiding method according to claim 1, wherein, the step of establishing matching line segments triple, specifically wraps
Include:
For the characteristic straight line in the three-view diagram of extraction, the characteristic straight line is established using LBD algorithms and describes son;
The matching line segments triple is established according to sub- nearest neighbouring rule is described.
3. unmanned plane barrier-avoiding method according to claim 1, line style obstacle is being estimated according to the matching line segments triple
Before the locus of thing, in addition to:
According to the outer parameter of three cameras, the camera matrix under structure normalization coordinate description;
The matrix of trifocal tensor is built according to the camera matrix;
Line-line-line three-view diagram geometrical constraint is determined according to the matrix of the trifocal tensor;
Matching line segments ternary wrong in the matching line segments triple is rejected according to the line-line-line three-view diagram geometrical constraint
Group, retain correct matching line segments triple.
4. unmanned plane barrier-avoiding method according to claim 3, wherein, the line-line-line three-view diagram geometrical constraint is:
Wherein,For the matching line segments triple,ForI-th of component (i=1,2,3),It is
Predicted value, TiFor the matrix of trifocal tensor;
Wherein,WhereinWithCamera matrix P is represented respectively2And P3I-th row;
Wherein, P1=[I | 0], It is jth phase
Posture of the machine in No. i-th camera coordinates system,Be kth camera relative to the translation vector of jth camera in No. i-th phase
Expression in machine coordinate system;And
The matching line segments triple is that the condition of correct matching line segments triple is:
Wherein, | | | | two norms of vector are represented, × represent two vectorial cross product operations.
5. unmanned plane barrier-avoiding method according to claim 1, the step of establishing Feature Points Matching triple, specifically includes:
To the angle point in the three-view diagram of extraction, feature point description is established using ORB algorithms;
Feature Points Matching triple is established according to nearest neighbouring rule.
6. unmanned plane barrier-avoiding method according to claim 1, non-linearity is being estimated according to the Feature Points Matching triple
Before the step of locus of barrier, in addition to:
According to the outer parameter of three cameras, the camera matrix under structure normalization coordinate description;
The matrix of trifocal tensor is built according to the camera matrix;
A point-line-three-view diagram geometrical relationship is determined according to the matrix of the trifocal tensor;
Feature Points Matching ternary wrong in Feature Points Matching triple is rejected according to a point-line-three-view diagram geometrical relationship
Group, retain correct Feature Points Matching triple.
7. unmanned plane barrier-avoiding method according to claim 6, wherein, a point-line-three-view diagram geometrical relationship is:
Wherein,It is first j-th of characteristic point of width viewThree components of homogeneous coordinates,It is
Predicted value, TiFor the matrix of trifocal tensor,WhereinWithCamera matrix is represented respectively
P2And P3I-th row;
Wherein, P1=[I | 0], It is jth number
Posture of the camera in No. i-th camera coordinates system,Be kth camera relative to the translation vector of jth camera at No. i-th
Expression in camera coordinates system;
It is to pass through in the 2nd width imagePut and perpendicular to the straight line to polar curve;And
The Feature Points Matching triple is that the condition of correct Feature Points Matching triple is:
Wherein, | | | | two norms of vector are represented, × represent two vectorial cross product operations.
8. unmanned plane barrier-avoiding method according to claim 7, according to the locus of the line style barrier and described non-
The locus of line style barrier, the step of establishing Obstacles distribution map, specifically include:
Place obstacles the resolution ratio of object space distribution map;
For the locus of the line style barrier of estimation, at interval of one between two end points of the line style barrier
Individual resolution ratio samples a point;
The locus of the point gathered on the line style barrier and the non-linearity barrier of the estimation is formed into three-dimensional point
Cloud;
The three-dimensional point cloud is converted into the Obstacles distribution map using OctoMap algorithms.
9. unmanned plane barrier-avoiding method according to claim 8, according to the Obstacles distribution map, acquisition can FOH
Domain, plan unmanned plane safe flight path the step of specifically include:
According to the Obstacles distribution map, optimal avoidance path is solved using OMPL tool storage rooms.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711078365.6A CN107885224A (en) | 2017-11-06 | 2017-11-06 | Unmanned plane barrier-avoiding method based on tri-item stereo vision |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711078365.6A CN107885224A (en) | 2017-11-06 | 2017-11-06 | Unmanned plane barrier-avoiding method based on tri-item stereo vision |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107885224A true CN107885224A (en) | 2018-04-06 |
Family
ID=61778788
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711078365.6A Pending CN107885224A (en) | 2017-11-06 | 2017-11-06 | Unmanned plane barrier-avoiding method based on tri-item stereo vision |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107885224A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108844538A (en) * | 2018-05-07 | 2018-11-20 | 中国人民解放军国防科技大学 | Unmanned aerial vehicle obstacle avoidance waypoint generation method based on vision/inertial navigation |
CN109446886A (en) * | 2018-09-07 | 2019-03-08 | 百度在线网络技术(北京)有限公司 | Obstacle detection method, device, equipment and storage medium based on unmanned vehicle |
CN109709554A (en) * | 2018-12-13 | 2019-05-03 | 广州极飞科技有限公司 | Operating equipment and its control method and device |
CN109753885A (en) * | 2018-12-14 | 2019-05-14 | 中国科学院深圳先进技术研究院 | A kind of object detection method, device and pedestrian detection method, system |
CN109917814A (en) * | 2019-04-19 | 2019-06-21 | 成都蔚来空间科技有限公司 | Unmanned plane operational method and system |
CN111781949A (en) * | 2020-07-03 | 2020-10-16 | 江苏大学 | Method for avoiding rod-shaped obstacle by unmanned aerial vehicle |
CN112393732A (en) * | 2020-11-24 | 2021-02-23 | 北京三快在线科技有限公司 | Unmanned aerial vehicle obstacle avoidance method and device, readable storage medium and electronic equipment |
CN112525106A (en) * | 2020-10-23 | 2021-03-19 | 清华大学 | Three-phase machine cooperative laser-based 3D detection method and device |
CN114047787A (en) * | 2021-12-23 | 2022-02-15 | 普宙科技(深圳)有限公司 | Unmanned aerial vehicle obstacle avoidance system, method, equipment and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101750029A (en) * | 2008-12-10 | 2010-06-23 | 中国科学院沈阳自动化研究所 | Characteristic point three-dimensional reconstruction method based on trifocal tensor |
WO2014072511A1 (en) * | 2012-11-09 | 2014-05-15 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Determination of rectifying homographies for a camera array |
CN104732577A (en) * | 2015-03-10 | 2015-06-24 | 山东科技大学 | Building texture extraction method based on UAV low-altitude aerial survey system |
CN105892489A (en) * | 2016-05-24 | 2016-08-24 | 国网山东省电力公司电力科学研究院 | Multi-sensor fusion-based autonomous obstacle avoidance unmanned aerial vehicle system and control method |
CN106530355A (en) * | 2016-10-31 | 2017-03-22 | 河海大学 | Multi-view camera synchronous data acquisition method based on optic-fiber Ethernet |
CN106909877A (en) * | 2016-12-13 | 2017-06-30 | 浙江大学 | A kind of vision based on dotted line comprehensive characteristics builds figure and localization method simultaneously |
-
2017
- 2017-11-06 CN CN201711078365.6A patent/CN107885224A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101750029A (en) * | 2008-12-10 | 2010-06-23 | 中国科学院沈阳自动化研究所 | Characteristic point three-dimensional reconstruction method based on trifocal tensor |
WO2014072511A1 (en) * | 2012-11-09 | 2014-05-15 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Determination of rectifying homographies for a camera array |
CN104732577A (en) * | 2015-03-10 | 2015-06-24 | 山东科技大学 | Building texture extraction method based on UAV low-altitude aerial survey system |
CN105892489A (en) * | 2016-05-24 | 2016-08-24 | 国网山东省电力公司电力科学研究院 | Multi-sensor fusion-based autonomous obstacle avoidance unmanned aerial vehicle system and control method |
CN106530355A (en) * | 2016-10-31 | 2017-03-22 | 河海大学 | Multi-view camera synchronous data acquisition method based on optic-fiber Ethernet |
CN106909877A (en) * | 2016-12-13 | 2017-06-30 | 浙江大学 | A kind of vision based on dotted line comprehensive characteristics builds figure and localization method simultaneously |
Non-Patent Citations (8)
Title |
---|
RICHARD HARTLEY ET AL.: "《计算机视觉中的多视图几何》", 30 August 2002, 安徽大学出版社 * |
卢俊: "基于无序多视影像的三维重建关键技术研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
吴伟仁 等: "《深空探测器自主导航原理与技术》", 31 May 2011, 中国宇航出版社 * |
孙水发 等: "《3D打印逆向建模技术及应用》", 31 May 2016, 南京师范大学出版社 * |
欧阳自远: "《中国探月工程》", 31 October 2014, 浙江科学技术出版社 * |
程磊: "《移动机器人系统及其协调控制》", 31 March 2014, 华中科技大学出版社 * |
胡永红 等: "《多肽类抗生素安来霉素生产工艺学》", 30 April 2016, 江苏凤凰科学技术出版社 * |
谢晓佳: "基于点线综合特征的双目视觉SLAM方法", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108844538B (en) * | 2018-05-07 | 2021-01-19 | 中国人民解放军国防科技大学 | Unmanned aerial vehicle obstacle avoidance waypoint generation method based on vision/inertial navigation |
CN108844538A (en) * | 2018-05-07 | 2018-11-20 | 中国人民解放军国防科技大学 | Unmanned aerial vehicle obstacle avoidance waypoint generation method based on vision/inertial navigation |
CN109446886A (en) * | 2018-09-07 | 2019-03-08 | 百度在线网络技术(北京)有限公司 | Obstacle detection method, device, equipment and storage medium based on unmanned vehicle |
CN109446886B (en) * | 2018-09-07 | 2020-08-25 | 百度在线网络技术(北京)有限公司 | Obstacle detection method, device, equipment and storage medium based on unmanned vehicle |
US11043002B2 (en) | 2018-09-07 | 2021-06-22 | Baidu Online Network Technology Co., Ltd. | Obstacle detecting method and obstacle detecting apparatus based on unmanned vehicle, and device, and storage medium |
CN109709554A (en) * | 2018-12-13 | 2019-05-03 | 广州极飞科技有限公司 | Operating equipment and its control method and device |
CN109753885A (en) * | 2018-12-14 | 2019-05-14 | 中国科学院深圳先进技术研究院 | A kind of object detection method, device and pedestrian detection method, system |
CN109753885B (en) * | 2018-12-14 | 2020-10-16 | 中国科学院深圳先进技术研究院 | Target detection method and device and pedestrian detection method and system |
CN109917814A (en) * | 2019-04-19 | 2019-06-21 | 成都蔚来空间科技有限公司 | Unmanned plane operational method and system |
CN111781949A (en) * | 2020-07-03 | 2020-10-16 | 江苏大学 | Method for avoiding rod-shaped obstacle by unmanned aerial vehicle |
CN111781949B (en) * | 2020-07-03 | 2022-11-18 | 江苏大学 | Method for avoiding rod-shaped obstacle by unmanned aerial vehicle |
CN112525106A (en) * | 2020-10-23 | 2021-03-19 | 清华大学 | Three-phase machine cooperative laser-based 3D detection method and device |
CN112525106B (en) * | 2020-10-23 | 2022-08-26 | 清华大学 | Three-phase machine cooperative laser-based 3D detection method and device |
CN112393732A (en) * | 2020-11-24 | 2021-02-23 | 北京三快在线科技有限公司 | Unmanned aerial vehicle obstacle avoidance method and device, readable storage medium and electronic equipment |
CN112393732B (en) * | 2020-11-24 | 2023-01-31 | 北京三快在线科技有限公司 | Unmanned aerial vehicle obstacle avoidance method and device, readable storage medium and electronic equipment |
CN114047787A (en) * | 2021-12-23 | 2022-02-15 | 普宙科技(深圳)有限公司 | Unmanned aerial vehicle obstacle avoidance system, method, equipment and storage medium |
CN114047787B (en) * | 2021-12-23 | 2024-04-05 | 普宙科技有限公司 | Unmanned aerial vehicle obstacle avoidance system, unmanned aerial vehicle obstacle avoidance method, unmanned aerial vehicle obstacle avoidance equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107885224A (en) | Unmanned plane barrier-avoiding method based on tri-item stereo vision | |
CN110415342B (en) | Three-dimensional point cloud reconstruction device and method based on multi-fusion sensor | |
CN105866790B (en) | A kind of laser radar obstacle recognition method and system considering lasing intensity | |
CN105573318B (en) | environment construction method based on probability analysis | |
CN108981672A (en) | Hatch door real-time location method based on monocular robot in conjunction with distance measuring sensor | |
CN111260773B (en) | Three-dimensional reconstruction method, detection method and detection system for small obstacle | |
CN106679634B (en) | A kind of space non-cooperative target pose measuring method based on stereoscopic vision | |
CN103093459B (en) | Utilize the method that airborne LiDAR point cloud data assisted image mates | |
CN110688905B (en) | Three-dimensional object detection and tracking method based on key frame | |
CN107560592B (en) | Precise distance measurement method for photoelectric tracker linkage target | |
CN108107444A (en) | Substation's method for recognizing impurities based on laser data | |
Cheng et al. | Building boundary extraction from high resolution imagery and lidar data | |
CN111260715B (en) | Depth map processing method, small obstacle detection method and system | |
CN113469178B (en) | Power meter identification method based on deep learning | |
CN114764885A (en) | Obstacle detection method and device, computer-readable storage medium and processor | |
Majdik et al. | Micro air vehicle localization and position tracking from textured 3d cadastral models | |
Kochi et al. | 3D modeling of architecture by edge-matching and integrating the point clouds of laser scanner and those of digital camera | |
Lim et al. | MSDPN: Monocular depth prediction with partial laser observation using multi-stage neural networks | |
Kovacs et al. | Edge detection in discretized range images | |
Li et al. | Real time obstacle estimation based on dense stereo vision for robotic lawn mowers | |
CN112197773A (en) | Visual and laser positioning mapping method based on plane information | |
John et al. | Sensor fusion and registration of lidar and stereo camera without calibration objects | |
Ru et al. | Material Location and Stacking Detection Method Based on LIDAR and Camera Fusion | |
Yuen et al. | Natural landmark based localisation system using panoramic images | |
Zhu et al. | Target Measurement Method Based on Sparse Disparity for Live Power Lines Maintaining Robot |
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: 20180406 |
|
RJ01 | Rejection of invention patent application after publication |