CN110335308A - The binocular vision speedometer calculation method examined based on disparity constraint and two-way annular - Google Patents
The binocular vision speedometer calculation method examined based on disparity constraint and two-way annular Download PDFInfo
- Publication number
- CN110335308A CN110335308A CN201910578572.0A CN201910578572A CN110335308A CN 110335308 A CN110335308 A CN 110335308A CN 201910578572 A CN201910578572 A CN 201910578572A CN 110335308 A CN110335308 A CN 110335308A
- Authority
- CN
- China
- Prior art keywords
- characteristic point
- moment
- image
- point
- examined
- 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
- 238000004364 calculation method Methods 0.000 title claims abstract description 34
- 238000000034 method Methods 0.000 claims abstract description 113
- 230000003287 optical effect Effects 0.000 claims abstract description 28
- 230000003044 adaptive effect Effects 0.000 claims abstract description 26
- 238000001514 detection method Methods 0.000 claims abstract description 15
- 238000000605 extraction Methods 0.000 claims description 13
- 238000012216 screening Methods 0.000 claims description 13
- 238000007689 inspection Methods 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 4
- 238000000354 decomposition reaction Methods 0.000 claims description 4
- 238000012217 deletion Methods 0.000 claims description 3
- 230000037430 deletion Effects 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 8
- 230000008569 process Effects 0.000 description 9
- 230000006870 function Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 1
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000026676 system process Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/20—Instruments for performing navigational calculations
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C22/00—Measuring distance traversed on the ground by vehicles, persons, animals or other moving solid bodies, e.g. using odometers, using pedometers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
- G06T7/75—Determining position or orientation of objects or cameras using feature-based methods involving models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/757—Matching configurations of points or features
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Remote Sensing (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Radar, Positioning & Navigation (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Databases & Information Systems (AREA)
- Automation & Control Theory (AREA)
- Image Analysis (AREA)
- Length Measuring Devices By Optical Means (AREA)
- Image Processing (AREA)
Abstract
The invention belongs to field of locating technology, more particularly to the binocular vision speedometer calculation method examined based on disparity constraint and two-way annular, it is intended that the accuracy rate of solution existing binocular vision speedometer feature association is low, and the low problem low with real-time of bring positioning accuracy.The method of the present invention includes: to carry out Image Acquisition by the binocular camera being loaded on mobile vehicle；Newly-increased feature point is obtained by Shi-Tomasi angular-point detection method to setting image and obtains new set of characteristic points；Characteristic point is tracked using KLT optical flow method, and the method by combining disparity constraint and adaptive two-way annular to examine establishes feature association；Estimated based on feature association as a result, obtaining initial pose using PNP method, and by characteristic point trigonometric ratio；Using light-stream adjustment, the final pose of the optimal estimation of pose is obtained by minimum re-projection error.The present invention improves the quality and efficiency of signature tracking, improves the positioning accuracy and real-time of binocular vision speedometer.
Description
Technical field
The invention belongs to field of locating technology, and in particular to a kind of binocular vision examined based on disparity constraint with two-way annular
Feel odometer calculation method.
Background technique
Intelligent vehicle receives each research institution, the world and colleges and universities as most active a part is developed in intelligent transportation system
Pay close attention to.Location technology is one of the key technology of intelligent vehicle safety movement, is played in automatic Pilot field important
Effect.
Binocular vision speedometer is a kind of practical vision positioning technology, and the pose of vehicle can be effectively estimated.With figure
As the development of processing technique and calculation power, binocular vision speedometer is gradually applied in embedded system, and becomes the one of automatic Pilot
A important component.Its advantage is that at low cost, energy consumption is small, easy for installation, good portability, anti-electromagnetic interference capability are strong
Deng.
The key step of binocular vision speedometer is divided into feature association and pose estimation.Feature association be based on feature extraction and
Tracking is established, and pose estimation is then based on feature association as a result, being obtained by minimizing re-projection error.Therefore, it establishes effective
Feature association be highly important.Method general at present is examined using front and back and left and right is examined and obtained to optical flow tracking
Feature association is tested, but this method still includes more exterior point, there are low-quality feature point tracking, precision and in real time
Property needs to be further increased.
Summary of the invention
In order to solve the above problem in the prior art, in order to solve the standard of existing binocular vision speedometer feature association
True rate is low, and the low problem low with real-time of bring positioning accuracy, first aspect present invention, proposes a kind of based on disparity constraint
The binocular vision speedometer calculation method examined with two-way annular, this method comprises:
Step S100 carries out left hand view according to the frequency acquisition of setting by the binocular camera being loaded on mobile vehicle
The acquisition of picture, image right, obtains the corresponding left-side images of t-1, t moment, image right；
Step S200 carries out Shi-Tomasi Corner Detection to the setting image at t-1 moment obtained in step S100, mentions
It is new to construct the setting image for the characteristic point of the setting image for taking newly-increased feature point set, and obtaining in conjunction with the t-1 moment
Set of characteristic points；
Step S300, based on the new set of characteristic points of the setting image, using KLT optical flow method, by disparity constraint and
Adaptive two-way annular is examined, and obtains the corresponding set of characteristic points of other images of t-1, t moment respectively, and carry out feature pass
Connection；
Step S400, based in step S300 feature association as a result, obtaining the movement using PNP position and orientation estimation method
The initial pose of carrier is estimated, and carries out three to characteristic point each in t moment left-side images, image right by binocular vision method
Angling obtains the corresponding three dimensional space coordinate of each characteristic point；
Step S500, result, the initial pose estimation, t moment left-side images, the right side based on step S300 feature association
The three dimensional space coordinate of each characteristic point and two dimensional image coordinate in the image of side, using light-stream adjustment, by minimizing re-projection
Error obtains the maximal possibility estimation of the pose of the mobile vehicle, obtains final pose.
In some preferred embodiments, left-side images collected, image right pass through OpenCV in step S100
Library function is corrected.
In some preferred embodiments, " characteristic point for the setting image that the t-1 moment obtains ", acquisition methods
Are as follows:
At the t-1 moment, the binocular image at t-2, t-1 moment is obtained using the method tracking of step S200, step S300
The t-1 moment sets the characteristic point of image.
In some preferred embodiments, " newly-increased feature point set is extracted " in step S200, method are as follows:
The characteristic point of the setting image at t-1 moment is extracted by Shi-Tomasi angular-point detection method, and will be fallen into
The feature point deletion of the primitive character point setting range neighborhood of the setting image, obtains newly-increased feature point；The setting image
Primitive character be the characteristic point of the setting image obtained at the t-1 moment.
In some preferred embodiments, the left-side images for setting image and being acquired as t moment.
In some preferred embodiments, step S300 " obtains the corresponding feature of other images of t-1, t moment respectively
Point set, and carry out feature association ", method are as follows:
By the left-side images at t-1 momentCharacteristic pointIt is tracked to obtain on a t moment left side by KLT optical flow method
Side imageCharacteristic pointThe characteristic point that screening t-1 moment left-side images are examined by front and back, and establishWithSpy
Sign association；
By t moment left-side imagesCharacteristic pointIt is tracked to obtain in t moment image right by KLT optical flow method
On characteristic pointThe characteristic point with disparity constraint is examined by left and right, and is establishedWithFeature association；
It is by t moment image rightOn characteristic pointIt is tracked to obtain on the right side of the t-1 moment by KLT optical flow method
ImageOn characteristic pointScreen t moment image rightPass through the adaptive rear preceding characteristic point examined.
In some preferred embodiments, " the corresponding spy of other images of t-1, t moment are obtained respectively in step S300
Levy point set ", it, will if t-1 is initial timeCharacteristic pointIt is tracked to obtain by KLT optical flow methodOn feature
PointScreening obtains the characteristic point by left and right inspection and disparity constraint.
In some preferred embodiments, " screening obtains the characteristic point by left and right inspection and disparity constraint ", side
Method are as follows:
It obtainsWithY-coordinate difference be less than given threshold ρ1Characteristic point, and by characteristic pointTraceback obtains
?On characteristic pointIfWithDistance be less than given threshold δ1, then examined by left and right.
In some preferred embodiments, " screening t-1 moment left-side images in step S301In pass through front and back examine
The characteristic point tested ", method are as follows:
By characteristic pointTraceback obtainsOn characteristic point
IfWithDistance be less than given threshold δ2, then as the characteristic point examined by front and back.
In some preferred embodiments, " screening t moment left-side images in step S302It is examined and is regarded by left and right
The characteristic point of difference constraint ", method are as follows:
If characteristic pointWithY-coordinate difference be less than given threshold ρ1, then retain this feature point and execute with lower section
Otherwise case deletes this feature point；
By characteristic pointTraceback obtainsOn characteristic point
IfWithDistance be less than given threshold δ3, then as the characteristic point examined by left and right.
In some preferred embodiments, " screening t moment image right in step S303Pass through adaptive rear preceding inspection
The characteristic point tested ", method are as follows:
By characteristic pointTraceback obtainsOn characteristic point
IfWithDistance be less than adaptive threshold δ4, then as the characteristic point by adaptive rear preceding inspection；
Wherein, adaptive threshold δ4Calculation method are as follows:
Wherein, ρ and ε is parameter, and loss is the characteristic point of t-1 moment image right feature point tracking t moment left-side images
When with the characteristic point number lost, maxtrack is the maximum tracking number of current signature point.
In some preferred embodiments, " maximum of pose is obtained by minimizing re-projection error in step S500
Possibility predication ", method are as follows:
According to feature association as a result, to each characteristic point building space re-projection error and time weight on each picture frame
Projection error；
The maximal possibility estimation of the pose in sliding window is obtained using light-stream adjustment.
In some preferred embodiments, further include step S600 after step S500:
Based on the final pose at moment each before t moment, t moment is obtained by marginalisation method and Shu Er decomposition method
Priori item rp, and by the priori item rpPrior uncertainty as maximal possibility estimation in step S500.
A kind of the second aspect of the present invention, it is also proposed that binocular vision mileage examined based on disparity constraint with two-way annular
Computing system is counted, which includes binocular image acquiring unit, initial characteristics point extraction unit, feature point extraction and feature association
Unit, initial pose estimation unit, maximal possibility estimation unit；
The binocular image acquiring unit, is configured to the binocular camera by being loaded on mobile vehicle, according to setting
Frequency acquisition carries out the acquisition of left-side images, image right, obtains the corresponding left-side images of t-1, t moment, image right；
The initial characteristics point extraction unit, was configured to the t-1 moment obtained in the binocular image acquiring unit
It sets image and carries out Shi-Tomasi Corner Detection, the setting extracting newly-increased feature point set, and obtaining in conjunction with the t-1 moment
The characteristic point of image constructs the new set of characteristic points of the setting image；
The feature point extraction and feature association unit are configured to make the set of characteristic points new based on the setting image,
Using KLT optical flow method, is examined by disparity constraint and adaptive two-way annular, obtain other images pair of t-1, t moment respectively
The set of characteristic points answered, and carry out feature association；
The initial pose estimation unit is configured to the feature association as a result, obtaining using PNP position and orientation estimation method
The initial pose estimation of the mobile vehicle is obtained, and by binocular vision method to each spy in t moment left-side images, image right
Sign point carries out trigonometric ratio, obtains the corresponding three dimensional space coordinate of each characteristic point；
The maximal possibility estimation unit is configured to the feature association, the initial pose estimation, a t moment left side
The three dimensional space coordinate of each characteristic point and two dimensional image coordinate pass through minimum using light-stream adjustment in side image, image right
Change the maximal possibility estimation that re-projection error obtains the pose of the mobile vehicle, obtains final pose.
The third aspect of the present invention, a kind of storage device, wherein being stored with a plurality of program, which is characterized in that described program
It is above-mentioned based on disparity constraint and the two-way annular binocular vision speedometer examined to realize using being loaded by processor and being executed
Calculation method.
The fourth aspect of the present invention, a kind of processing setting, including processor, storage device；Processor is adapted for carrying out each
Program；Storage device is suitable for storing a plurality of program；It is characterized in that, described program is suitable for being loaded and being held by processor
Row is to realize the above-mentioned binocular vision speedometer calculation method examined based on disparity constraint with two-way annular.
Beneficial effects of the present invention:
Present invention combination disparity constraint, can effectively purify characteristic point, avoid low-quality characteristic point with
Track, to improve the quality and efficiency of signature tracking；The invention proposes adaptive two-way annulars to examine, and further eliminates outer
Point improves the accuracy rate of feature association, and can be adaptively adjusted feature for the characteristic point variation in different motion
The quality and quantity of point, to improve the positioning accuracy and real-time of binocular vision speedometer.
Detailed description of the invention
By reading a detailed description of non-restrictive embodiments in the light of the attached drawings below, the application's is other
Feature, objects and advantages will become more apparent upon:
Fig. 1 is that the binocular vision speedometer examined based on disparity constraint and two-way annular of an embodiment of the present invention is calculated
Method flow schematic diagram；
Fig. 2 is that the binocular vision speedometer examined based on disparity constraint and two-way annular of an embodiment of the present invention is calculated
Method disparity constraint schematic diagram；
Fig. 3 is that the binocular vision speedometer examined based on disparity constraint and two-way annular of an embodiment of the present invention is calculated
The adaptive two-way annular of method examines schematic diagram.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with attached drawing to the embodiment of the present invention
In technical solution be clearly and completely described, it is clear that described embodiments are some of the embodiments of the present invention, without
It is whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not before making creative work
Every other embodiment obtained is put, shall fall within the protection scope of the present invention.
The application is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is only used for explaining related invention, rather than the restriction to the invention.It also should be noted that in order to just
Part relevant to related invention is illustrated only in description, attached drawing.
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
Mutually combination.
The binocular vision speedometer calculation method examined based on disparity constraint and two-way annular of an embodiment of the present invention,
Specifically includes the following steps:
Step S100 carries out left hand view according to the frequency acquisition of setting by the binocular camera being loaded on mobile vehicle
The acquisition of picture, image right, obtains the corresponding left-side images of t-1, t moment, image right；
Step S200 carries out Shi-Tomasi Corner Detection to the setting image at t-1 moment obtained in step S100, mentions
It is new to construct the setting image for the characteristic point of the setting image for taking newly-increased feature point set, and obtaining in conjunction with the t-1 moment
Set of characteristic points；
Step S300, based on the new set of characteristic points of the setting image, using KLT optical flow method, by disparity constraint and
Adaptive two-way annular is examined, and obtains the corresponding set of characteristic points of other images of t-1, t moment respectively, and carry out feature pass
Connection；
Step S400, based in step S300 feature association as a result, obtaining the movement using PNP position and orientation estimation method
The initial pose of carrier is estimated, and carries out three to characteristic point each in t moment left-side images, image right by binocular vision method
Angling obtains the corresponding three dimensional space coordinate of each characteristic point；
Step S500, result, the initial pose estimation, t moment left-side images, the right side based on step S300 feature association
The three dimensional space coordinate of each characteristic point and two dimensional image coordinate in the image of side, using light-stream adjustment, by minimizing re-projection
Error obtains the maximal possibility estimation of the pose of the mobile vehicle, obtains final pose.In some preferred embodiments,
Further include step S600 after step S500:
Based on the final pose at moment each before t moment, t moment is obtained by marginalisation method and Shu Er decomposition method
Priori item rp, and by the priori item rpPrior uncertainty as maximal possibility estimation in step S500.
In order to more clearly to the present invention is based on the binocular vision speedometer calculating sides that disparity constraint and two-way annular are examined
Method is illustrated, with reference to the accompanying drawing to a kind of embodiment of our inventive method based on disparity constraint and two-way annular inspection
Each step carries out expansion detailed description in binocular vision speedometer calculation method.
The binocular vision speedometer calculation method examined based on disparity constraint and two-way annular of an embodiment of the present invention,
Include the steps that S100-S600 described below.
Step S100 carries out left hand view according to the frequency acquisition of setting by the binocular camera being loaded on mobile vehicle
The acquisition of picture, image right, obtains the corresponding left-side images of t-1, t moment, image right；
In the present embodiment, it before carrying out Image Acquisition, needs to demarcate the binocular camera assembled, calibration
Parameter includes parameter between the camera intrinsic parameter of binocular camera and camera；
Left-side images collected, image right are corrected by OpenCV library function.
Step S200 carries out Shi-Tomasi Corner Detection to the setting image at t-1 moment obtained in step S100, mentions
It is new to construct the setting image for the characteristic point of the setting image for taking newly-increased feature point set, and obtaining in conjunction with the t-1 moment
Set of characteristic points；
In the present embodiment, the quantity for the preset total characteristic point that can be set is 200, certainly in some other implementation
In example or other setting values, such as 100,150 etc..
The characteristic point for the setting image that the t-1 moment in this step obtains, acquisition methods are as follows:
At the t-1 moment, the binocular image at t-2, t-1 moment is obtained using the method tracking of step S200, step S300
The t-1 moment sets the characteristic point of image.
In this step, newly-increased feature point set, method are extracted are as follows:
The characteristic point of the setting image at t-1 moment is extracted by Shi-Tomasi angular-point detection method, and will be fallen into
The feature point deletion of the primitive character point setting range neighborhood of the setting image, obtains newly-increased feature point.Set the original of image
Beginning feature is the characteristic point of the setting image obtained at the t-1 moment.Primitive character point setting range neighborhood are as follows: be based on original spy
Sign point obtains the corresponding border circular areas of each primitive character point according to the radius of setting respectively, and the intersection of all border circular areas is original
Beginning characteristic point setting range neighborhood.
In the present embodiment, the image for carrying out Shi-Tomasi Corner Detection is the left-side images acquired at the t-1 moment, when
It so, can also be the image right of t moment acquisition in some other embodiment.
Shi-Tomasi Corner Detection determines characteristic point, matching characteristic by comparing the minimal eigenvalue of gradient matrix
When point, which introduce affine transformations, make characteristic point more accurate in frame matching, and exclude bad characteristic point.This step
Feature extracting method is on the books in the more document in this field, herein not reinflated detailed description.
Step S300, based on the new set of characteristic points of the setting image, using KLT optical flow method, by disparity constraint and
Adaptive two-way annular is examined, and obtains the corresponding set of characteristic points of other images of t-1, t moment respectively, and carry out feature pass
Connection.
Further include the steps that feature point tracking number counts in the present embodiment: in t moment pose acquisition process, setting figure
As characteristic point then tracks success, this feature point by disparity constraint and adaptive two-way annular inspection in new set of characteristic points
Maximum tracking number add 1, to realize the statistics of tracking number.For newly-increased characteristic point, initially tracking number is 0.
Optical flow method is used for the principle of target detection: assigning a velocity vector to each of image pixel, in this way
It is formed a motion vector field.In a certain particular moment, the point on image and the point on three-dimension object are corresponding one by one, in this way
Corresponding relation can be calculated by projecting.
1, the initial time that Image Acquisition is carried out in binocular camera, only obtains a pair of of image, i.e. t=1, and t-1 is not present,
At this point, screening the moment left-side imagesIn examined by left and right and the characteristic point of disparity constraint, and carry out left-side images, right side
The matching of image characteristic point.
It is in the present embodiment the specific scheme is that ifCharacteristic point can by left and right examine and disparity constraint, then retain
This feature point, and feature association is established, this feature point is otherwise deleted, method particularly includes:
Step 3001: willCharacteristic pointIt is tracked to obtain by KLT optical flow methodOn characteristic point
Step 3002, if characteristic pointWithY-coordinate difference be less than given threshold ρ1, then retain this feature point and go forward side by side
Otherwise row following steps delete this feature point and skip following steps；
Step 3003, by characteristic pointTraceback obtainsOn characteristic point
Step 3004, ifWithDistance be less than given threshold δ1, then examined by left and right.
2, each moment of binocular camera at the second acquisition moment and later, the binocular figure at former and later two available moment
Picture carries out feature association by following steps S301 to step S303 after the binocular image for obtaining t moment.
Step S301, by t-1 moment left-side imagesCharacteristic pointIt is tracked to obtain by KLT optical flow method
T moment left-side imagesOn characteristic pointScreen t-1 moment left-side imagesIn by front and back examine characteristic point and build
It is verticalWithFeature association.
In the implementation case this step the specific scheme is that ifCharacteristic point can be examined by front and back, then protect
This feature point is stayed, and establishes feature association, otherwise deletes this feature point, method particularly includes:
Step 3011, willCharacteristic pointIt is tracked to obtain by KLT optical flow methodOn characteristic point
Step 3012, by characteristic pointTraceback obtainsOn characteristic point
Step 3013, ifWithDistance be less than given threshold δ2, then examined by front and back.
Step S302, by t moment left-side imagesCharacteristic pointIt is tracked to obtain in t moment by KLT optical flow method
Image rightOn characteristic pointScreen t moment left-side imagesThe characteristic point with disparity constraint is examined by left and right, and is built
It is verticalWithFeature association.
In the present embodiment this step the specific scheme is that ifCharacteristic point can by left and right examine and parallax about
Beam then retains this feature point, and establishes feature association, otherwise deletes this feature point, method particularly includes:
Step 3021: willCharacteristic pointIt is tracked to obtain by KLT optical flow methodOn characteristic point
Step 3022: if characteristic pointWithY-coordinate difference be less than given threshold ρ1, then retain this feature point and go forward side by side
Otherwise row following steps delete this feature point and skip following steps；
Step 3023, by characteristic pointTraceback obtainsOn characteristic point
Step 3024, ifWithDistance be less than given threshold δ3, then examined by left and right.
It is illustrated in figure 2 the parallax for the binocular vision speedometer calculation method examined based on disparity constraint and two-way annular about
Beam schematic diagram, show in figure byCharacteristic pointIt is tracked to obtain by KLT optical flow methodOn
Characteristic pointDotted line is the allowed band [+ρ for showing y-coordinate in figure1,-ρ1],Middle characteristic pointRight side solid line indicates characteristic pointWith characteristic pointY-coordinate is equal.
Step S303, by t moment right part of flgOn characteristic pointIt is tracked to obtain in t-1 by KLT optical flow method
Carve image rightOn characteristic pointScreen t moment image rightPass through the adaptive rear preceding characteristic point examined.
In the present embodiment this step the specific scheme is that ifCharacteristic point can be by examining before after adaptive, then
Retain this feature point, otherwise delete this feature point, method particularly includes:
Step 3031, willCharacteristic pointIt is tracked to obtain by KLT optical flow methodOn characteristic point
Step 3032, by characteristic pointTraceback obtainsOn characteristic point
Step 3033, ifWithDistance be less than adaptive threshold δ4, then pass through adaptive rear preceding inspection；
Adaptive threshold δ4Equation such as formula (1) shown in:
Wherein, ρ and ε is parameter preset (value is ρ=0.23, ε=5 in the present embodiment), and loss is t moment left-side images
With the characteristic point number lost, (i.e. preset total characteristic point number subtracts t when the characteristic point of feature point tracking t-1 moment left-side images
Moment left-side images characteristic point successfully tracks the number of the characteristic point of t-1 moment left-side images, preset total spy in the present embodiment
Sign point number is that 200), maxtrack is the maximum tracking number of current signature point.In the present embodiment, each moment carries out position
When appearance calculates, two-way annular is examined since the left-side images of last moment.In some other embodiment, if two-way annular
It examines since the image right at t-1 moment, then loss is t moment image right feature point tracking t-1 moment right part of flg at this time
With the characteristic point number lost when the characteristic point of picture, in this case, two-way annular is examined since the image right of last moment.
It is illustrated in figure 3 adaptively double based on disparity constraint and the binocular vision speedometer calculation method of two-way annular inspection
Schematic diagram is examined to annular, is shown in figureTwo-way annular is carried out by KLT optical flow method to examine,
In, the solid arrow between every two image indicates that tracking is carried out by KLT optical flow method obtains characteristic point in arrow meaning image
Process, dotted arrow indicate traceback obtain the process of arrow meaning image characteristic point.In figureWithBetween
Solid arrow and dotted arrow are respectively the t-1 momentWithBetween by tracking and antitracking obtain individual features point
Process.
Step S400 is based on the feature association, the initial bit of the mobile vehicle is obtained using PNP position and orientation estimation method
Appearance estimation, and obtain carrying out trigonometric ratio to each characteristic point by binocular vision method, obtain the corresponding three-dimensional space of each characteristic point
Coordinate；
According to the two dimensional character of present frame point and corresponding three dimensional space coordinate, initial pose is obtained using PNP method and is estimated
Meter.The characteristic point that present frame is newly extracted carries out trigonometric ratio by the method for binocular vision, obtains corresponding three dimensional space coordinate,
It is resolved for subsequent pose.
Step S500, based on each in the feature association, the initial pose estimation, t moment left-side images, image right
The three dimensional space coordinate and two dimensional image coordinate of characteristic point obtain institute by minimizing re-projection error using light-stream adjustment
The maximal possibility estimation for stating the pose of mobile vehicle obtains final pose.
The pose that the embodiment of the present invention directly acquires is the pose of binocular camera left camera, passes through left camera and movement
The pose mapping relations of carrier, the pose of available mobile vehicle.In some other embodiment, if it is two-way annular examine from
The image right at t-1 moment starts, then is the pose of binocular camera right camera by the pose that the above method directly acquires, together
Sample passes through the pose mapping relations of right camera and mobile vehicle, the pose of available mobile vehicle.
According to feature association as a result, constructing re-projection error for each characteristic point on each picture frame.The re-projection
Error is by the picture frame after projecting to this feature point from the position of first time observation.Assuming that first observes spy
The picture frame for levying point l is i, then its re-projection error such as formula (2) on picture frame t is shown:
Wherein,It is equivalent toIndicate observation point of the characteristic point l on picture frame t；It is characteristic point l in image
The observation model of frame t, χ indicate pose；πcIt is pin-hole model, feature is projected into image coordinate system from camera coordinates system；T is phase
The homogeneous matrix of seat in the plane appearance, TiIndicate the correspondence pose of picture frame i, TtIndicate the correspondence pose of picture frame t.λlIndicate characteristic point l
Depth.
For binocular vision speedometer, this method constructs space re-projection error and time re-projection using this method simultaneously
Error.
It is estimated as initial value with initial pose, is estimated using the maximum likelihood that light-stream adjustment obtains the pose in sliding window
Meter can solve the equation using gauss-newton method as shown in formula (3).
Wherein, S is image measurement value set (i.e. the two dimensional image coordinate set of characteristic point),It is covariance matrix, χ*
It is optimal pose (the final pose resolved), χ is pose to be optimized, and n is moment serial number,It is characteristic point two dimensional image
Coordinate,For pin-hole model πc, rpFor priori item obtained in step S600.
Priori item r is increased in maximal possibility estimation in the present embodimentp, public certainly in some other embodiments
Formula (3) can also remove priori item rp。
In order to improve the present invention is based on disparity constraint and it is two-way annular examine binocular vision speedometer calculation method when
Effect property increases step S600 in some embodiments.
Step S600 is obtained based on the final pose at moment each before t moment by marginalisation method and Shu Er decomposition method
Take the priori item r of t momentp, and by the priori item rpPrior uncertainty as maximal possibility estimation in step S500.
By the step as system mode is continuously increased, system mode (final pose) number is reduced using marginalisation method
Amount guarantees that binocular vision speedometer being capable of real time execution to have the function that reduce computation complexity.
It is priori item r that marginalisation method, which is decomposed by Shu Er by part system condition conversion before,p, and by it from sliding
It is removed in window, provides prior information for the state of sliding window.
The pose and new picture frame pose that still retain after marginalisation constitute the pose in sliding window, and circulation uses light beam
The maximal possibility estimation of adjustment method acquisition pose.
It can be with specific reference to " Leutenegger S, Lynen S, Bosse M, et al.Keyframe- in this step
based visual-inertial odometry using nonlinear optimization[J].The
International Journal of Robotics Research, 2015,34 (3): 314-334. ", it is not reinflated herein
It is described in detail.
In the description of above-mentioned technical proposal, step S600 is intended merely to clearly after being placed on step S500 to technology
Scheme is described, rather than the restriction to sequence of steps, can be placed in front of step S500 in some embodiments.
The a kind of of second embodiment of the invention is calculated based on disparity constraint with the binocular vision speedometer of two-way annular inspection
System, the system include binocular image acquiring unit, initial characteristics point extraction unit, feature point extraction and feature association unit,
Initial pose estimation unit, maximal possibility estimation unit；
The binocular image acquiring unit, is configured to the binocular camera by being loaded on mobile vehicle, according to setting
Frequency acquisition carries out the acquisition of left-side images, image right, obtains t, t+1 moment corresponding left-side images, image right；
The initial characteristics point extraction unit, was configured to the t-1 moment obtained in the binocular image acquiring unit
It sets image and carries out Shi-Tomasi Corner Detection, the setting extracting newly-increased feature point set, and obtaining in conjunction with the t-1 moment
The characteristic point of image constructs the new set of characteristic points of the setting image；
The feature point extraction and feature association unit are configured to make the set of characteristic points new based on the setting image,
Using KLT optical flow method, is examined by disparity constraint and adaptive two-way annular, obtain other images pair at t, t+1 moment respectively
The set of characteristic points answered, and carry out feature association；
The initial pose estimation unit is configured to the feature association as a result, obtaining using PNP position and orientation estimation method
The initial pose estimation of the mobile vehicle is obtained, and by binocular vision method to each spy in t moment left-side images, image right
Sign point carries out trigonometric ratio, obtains the corresponding three dimensional space coordinate of each characteristic point；
The maximal possibility estimation unit is configured to the feature association, the initial pose estimation, a t moment left side
The three dimensional space coordinate of each characteristic point and two dimensional image coordinate in side image and image right, using light-stream adjustment, by most
Smallization re-projection error obtains the maximal possibility estimation of the pose of the mobile vehicle, obtains final pose.
Person of ordinary skill in the field can be understood that, for convenience and simplicity of description, foregoing description
The specific work process and related explanation of storage device, processing unit, can refer to corresponding processes in the foregoing method embodiment,
Details are not described herein.
It should be noted that the binocular vision mileage provided by the above embodiment examined based on disparity constraint with two-way annular
Count computing system, only the example of the division of the above functional modules, in practical applications, can according to need and incite somebody to action
Above-mentioned function distribution is completed by different functional modules, i.e., by the embodiment of the present invention module or step decompose again or
Combination, for example, the module of above-described embodiment can be merged into a module, can also be further split into multiple submodule, with
Complete all or part of function described above.For module involved in the embodiment of the present invention, the title of step, only
In order to distinguish modules or step, it is not intended as inappropriate limitation of the present invention.
Third of the present invention applies a kind of storage device of example, wherein being stored with a plurality of program, described program is suitable for by processor
It loads and executes to realize the above-mentioned binocular vision speedometer calculation method examined based on disparity constraint with two-way annular.
The present invention the 4th applies a kind of processing unit of example, including processor, storage device；Processor is adapted for carrying out each item
Program；Storage device is suitable for storing a plurality of program；Described program be suitable for loaded by processor and executed with realize it is above-mentioned based on
The binocular vision speedometer calculation method that disparity constraint and two-way annular are examined.
Person of ordinary skill in the field can be understood that, for convenience and simplicity of description, foregoing description
The specific work process and related explanation of storage device, processing unit, can refer to corresponding processes in the foregoing method embodiment,
Details are not described herein.
Those skilled in the art should be able to recognize that, mould described in conjunction with the examples disclosed in the embodiments of the present disclosure
Block, method and step, can be realized with electronic hardware, computer software, or a combination of the two, software module, method and step pair
The program answered can be placed in random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electric erasable and can compile
Any other form of storage well known in journey ROM, register, hard disk, moveable magnetic disc, CD-ROM or technical field is situated between
In matter.In order to clearly demonstrate the interchangeability of electronic hardware and software, in the above description according to function generally
Describe each exemplary composition and step.These functions are executed actually with electronic hardware or software mode, depend on technology
The specific application and design constraint of scheme.Those skilled in the art can carry out using distinct methods each specific application
Realize described function, but such implementation should not be considered as beyond the scope of the present invention.
Term " first ", " second " etc. are to be used to distinguish similar objects, rather than be used to describe or indicate specific suitable
Sequence or precedence.
Term " includes " or any other like term are intended to cover non-exclusive inclusion, so that including a system
Process, method, article or equipment/device of column element not only includes those elements, but also including being not explicitly listed
Other elements, or further include the intrinsic element of these process, method, article or equipment/devices.
So far, it has been combined preferred embodiment shown in the drawings and describes technical solution of the present invention, still, this field
Technical staff is it is easily understood that protection scope of the present invention is expressly not limited to these specific embodiments.Without departing from this
Under the premise of the principle of invention, those skilled in the art can make equivalent change or replacement to the relevant technologies feature, these
Technical solution after change or replacement will fall within the scope of protection of the present invention.
Claims (16)
1. a kind of binocular vision speedometer calculation method examined based on disparity constraint with two-way annular, which is characterized in that pose
Calculation method the following steps are included:
Step S100 carries out left-side images, the right side according to the frequency acquisition of setting by the binocular camera being loaded on mobile vehicle
The acquisition of side image obtains the corresponding left-side images of t-1, t moment, image right；
Step S200 carries out Shi-Tomasi Corner Detection to the setting image at t-1 moment obtained in step S100, extracts new
The characteristic point for the setting image for increasing set of characteristic points, and obtaining in conjunction with the t-1 moment constructs the new feature of the setting image
Point set；
Step S300, based on the new set of characteristic points of the setting image, using KLT optical flow method, by disparity constraint and adaptive
It answers two-way annular to examine, obtains the corresponding set of characteristic points of other images of t-1, t moment respectively, and carry out feature association；
Step S400, based in step S300 feature association as a result, obtaining the mobile vehicle using PNP position and orientation estimation method
Initial pose estimation, and by binocular vision method to characteristic point each in t moment left-side images and image right carry out triangle
Change, obtains the corresponding three dimensional space coordinate of each characteristic point；
Step S500, result, the initial pose estimation, t moment left-side images, right part of flg based on step S300 feature association
The three dimensional space coordinate and two dimensional image coordinate of each characteristic point as in, using light-stream adjustment, by minimizing re-projection error
The maximal possibility estimation of the pose of the mobile vehicle is obtained, final pose is obtained.
2. the binocular vision speedometer calculation method according to claim 1 examined based on disparity constraint with two-way annular,
It is characterized in that, left-side images collected, image right are corrected by OpenCV library function in step S100.
3. the binocular vision speedometer calculation method according to claim 1 examined based on disparity constraint with two-way annular,
It is characterized in that, " characteristic point for the setting image that the t-1 moment obtains ", acquisition methods in step S200 are as follows:
At the t-1 moment, the binocular image at t-2, t-1 moment, the t-1 obtained using the method tracking of step S200, step S300
The characteristic point of moment setting image.
4. the binocular vision speedometer calculation method according to claim 2 examined based on disparity constraint with two-way annular,
It is characterized in that, " newly-increased feature point set is extracted " in step S200, method are as follows:
The characteristic point of the setting image at t-1 moment is extracted by Shi-Tomasi angular-point detection method, and will be fallen into described
The feature point deletion for setting the primitive character point setting range neighborhood of image, obtains newly-increased feature point；The original of the setting image
Beginning feature is the characteristic point of the setting image obtained at the t-1 moment.
5. according to claim 1-4 based on the binocular vision speedometer that disparity constraint is examined with two-way annular
Calculation method, which is characterized in that the left-side images for setting image and being acquired as t moment.
6. the binocular vision speedometer calculation method according to claim 5 examined based on disparity constraint with two-way annular,
It is characterized in that, step S300 " obtains the corresponding set of characteristic points of other images of t-1, t moment respectively, and carries out feature pass
Connection ", method are as follows:
Step S301, by t-1 left-side images left imageCharacteristic pointIt is tracked to obtain in t by KLT optical flow method
Moment left imageCharacteristic pointThe characteristic point that screening t-1 moment left-side images are examined by front and back, and establishWithFeature association；
Step S302, by t moment left-side imagesCharacteristic pointIt is tracked to obtain on the right side of t moment by KLT optical flow method
ImageOn characteristic pointThe characteristic point with disparity constraint is examined by left and right, and is establishedWithFeature association；
T moment image right is by step S303On characteristic pointIt is tracked to obtain in t-1 by KLT optical flow method
Carve image rightOn characteristic pointScreen t moment image rightPass through the adaptive rear preceding characteristic point examined.
7. the binocular vision speedometer calculation method examined according to claim 5 based on disparity constraint and two-way annular,
It is characterized in that, " the corresponding set of characteristic points of other images of t-1, t moment is obtained respectively " in step S300, if t is initial
It carves, it willCharacteristic pointIt is tracked to obtain by KLT optical flow methodOn characteristic pointScreening, which obtains, passes through left and right
Examine the characteristic point with disparity constraint.
8. the binocular vision speedometer calculation method examined according to claim 7 based on disparity constraint and two-way annular,
It is characterized in that, " screening obtains the characteristic point by left and right inspection and disparity constraint ", method are as follows:
It obtainsWithY-coordinate difference be less than given threshold ρ1Characteristic point, and by characteristic pointTraceback obtains
On characteristic pointIfWithDistance be less than given threshold δ1, then examined by left and right.
9. the binocular vision speedometer calculation method examined according to claim 6 based on disparity constraint and two-way annular,
It is characterized in that, " screening t-1 moment left-side images in step S301In by front and back examine characteristic point ", method
Are as follows:
By characteristic pointTraceback obtainsOn characteristic point
IfWithDistance be less than given threshold δ2, then as the characteristic point examined by front and back.
10. the binocular vision speedometer calculation method according to claim 6 examined based on disparity constraint with two-way annular,
It is characterized in that, " screening t moment left-side images in step S302The characteristic point with disparity constraint is examined by left and right ", side
Method are as follows:
Step S3021, if characteristic pointWithY-coordinate difference be less than given threshold ρ1, then retain this feature point and execute
Otherwise step S3022 deletes this feature point；
Step S3022, by characteristic pointTraceback obtainsOn characteristic point
Step S3023, ifWithDistance be less than given threshold δ3, then as the characteristic point by left and right detection.
11. the binocular vision speedometer calculation method according to claim 6 examined based on disparity constraint with two-way annular,
It is characterized in that, " screening t moment image right in step S303Pass through the adaptive rear preceding characteristic point examined ", method
Are as follows:
By characteristic pointTraceback obtainsOn characteristic point
IfWithDistance be less than adaptive threshold δ4, then as the characteristic point by adaptive rear preceding inspection；
Wherein, adaptive threshold δ4Calculation method are as follows:
Wherein, ρ and ε is parameter preset, and loss is the characteristic point of t-1 moment image right feature point tracking t moment left-side images
When with the characteristic point number lost, maxtrack is the maximum tracking number of current signature point.
12. the binocular vision speedometer according to claim 1-4 examined based on disparity constraint with two-way annular
Calculation method, which is characterized in that in step S500 " maximal possibility estimation of pose is obtained by minimizing re-projection error ",
Method are as follows:
Step S501, according to feature association as a result, on each picture frame each characteristic point building space re-projection error and
Time re-projection error；
Step S502 obtains the maximal possibility estimation of pose in sliding window using light-stream adjustment.
13. the binocular vision speedometer according to claim 1-4 examined based on disparity constraint with two-way annular
Calculation method, which is characterized in that further include step S600 after step S500:
Based on the final pose at moment each before t moment, the priori of t moment is obtained by marginalisation method and Shu Er decomposition method
Item rp, and by the priori item rpPrior uncertainty as maximal possibility estimation in step S500.
14. a kind of binocular vision speedometer computing system examined based on disparity constraint with two-way annular, which is characterized in that this is
System includes that binocular image acquiring unit, initial characteristics point extraction unit, feature point extraction and feature association unit, initial pose are estimated
Count unit, maximal possibility estimation unit；
The binocular image acquiring unit, is configured to the binocular camera by being loaded on mobile vehicle, according to the acquisition of setting
Frequency carries out the acquisition of left-side images, image right, obtains t, t+1 moment corresponding left-side images, image right；
The initial characteristics point extraction unit, is configured to one to the t-1 moment obtained in the binocular image acquiring unit
Image carries out Shi-Tomasi Corner Detection, the setting image for extracting newly-increased feature point set, and obtaining in conjunction with the t-1 moment
Characteristic point, construct the new set of characteristic points of the setting image；
The feature point extraction and feature association unit are configured to make the set of characteristic points new based on the setting image, use
KLT optical flow method is examined by disparity constraint and adaptive two-way annular, other images for obtaining t, t+1 moment respectively are corresponding
Set of characteristic points, and carry out feature association；
The initial pose estimation unit is configured to the feature association as a result, obtaining institute using PNP position and orientation estimation method
The initial pose estimation of mobile vehicle is stated, and by binocular vision method to each characteristic point in t moment left-side images, image right
Trigonometric ratio is carried out, the corresponding three dimensional space coordinate of each characteristic point is obtained；
The maximal possibility estimation unit is configured to the feature association, the initial pose estimation, t moment left hand view
The three dimensional space coordinate of each characteristic point and two dimensional image coordinate pass through minimum using light-stream adjustment in picture and image right
Re-projection error obtains the maximal possibility estimation of the pose of the mobile vehicle.
15. a kind of storage device, wherein being stored with a plurality of program, which is characterized in that described program is applied by processor load simultaneously
It executes to realize the described in any item binocular vision speedometers examined based on disparity constraint with two-way annular of claim 1-13
Calculation method.
16. a kind of processing setting, including processor, storage device；Processor is adapted for carrying out each program；Storage device is fitted
For storing a plurality of program；It is characterized in that, described program is suitable for being loaded by processor and being executed to realize claim 1-
13 described in any item binocular vision speedometer calculation methods examined based on disparity constraint with two-way annular.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910578572.0A CN110335308B (en) | 2019-06-28 | 2019-06-28 | Binocular vision odometer calculation method based on parallax constraint and bidirectional annular inspection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910578572.0A CN110335308B (en) | 2019-06-28 | 2019-06-28 | Binocular vision odometer calculation method based on parallax constraint and bidirectional annular inspection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110335308A true CN110335308A (en) | 2019-10-15 |
| CN110335308B CN110335308B (en) | 2021-07-30 |
Family
ID=68144566
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910578572.0A Active CN110335308B (en) | 2019-06-28 | 2019-06-28 | Binocular vision odometer calculation method based on parallax constraint and bidirectional annular inspection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110335308B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111862150A (en) * | 2020-06-19 | 2020-10-30 | 杭州易现先进科技有限公司 | Image tracking method and device, AR device and computer device |
| CN112330589A (en) * | 2020-09-18 | 2021-02-05 | 北京沃东天骏信息技术有限公司 | Method and device for estimating pose and computer readable storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130182894A1 (en) * | 2012-01-18 | 2013-07-18 | Samsung Electronics Co., Ltd. | Method and apparatus for camera tracking |
| CN107025668A (en) * | 2017-03-30 | 2017-08-08 | 华南理工大学 | A kind of design method of the visual odometry based on depth camera |
| CN108519102A (en) * | 2018-03-26 | 2018-09-11 | 东南大学 | A kind of binocular vision speedometer calculation method based on reprojection |
-
2019
- 2019-06-28 CN CN201910578572.0A patent/CN110335308B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130182894A1 (en) * | 2012-01-18 | 2013-07-18 | Samsung Electronics Co., Ltd. | Method and apparatus for camera tracking |
| CN107025668A (en) * | 2017-03-30 | 2017-08-08 | 华南理工大学 | A kind of design method of the visual odometry based on depth camera |
| CN108519102A (en) * | 2018-03-26 | 2018-09-11 | 东南大学 | A kind of binocular vision speedometer calculation method based on reprojection |
Non-Patent Citations (2)
| Title |
|---|
| HOWARD A.等: "《Real-time stereo visual odometry for autonomous ground vehicles》", 《2008 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS》 * |
| 赵彤 等: "《三阶段局部双目光束法平差视觉里程计》", 《光电工程》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111862150A (en) * | 2020-06-19 | 2020-10-30 | 杭州易现先进科技有限公司 | Image tracking method and device, AR device and computer device |
| CN111862150B (en) * | 2020-06-19 | 2024-06-14 | 杭州易现先进科技有限公司 | Image tracking method, device, AR equipment and computer equipment |
| CN112330589A (en) * | 2020-09-18 | 2021-02-05 | 北京沃东天骏信息技术有限公司 | Method and device for estimating pose and computer readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110335308B (en) | 2021-07-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP3557982B2 (en) | Optical flow estimation method | |
| CN113593017A (en) | Method, device and equipment for constructing surface three-dimensional model of strip mine and storage medium | |
| CN109903372A (en) | Depth map super-resolution complementing method and high quality three-dimensional rebuilding method and system | |
| EP3293700B1 (en) | 3d reconstruction for vehicle | |
| CN112950696A (en) | Navigation map generation method and generation device and electronic equipment | |
| Gibson et al. | Robust optical flow estimation based on a sparse motion trajectory set | |
| CN112767478B (en) | Appearance guidance-based six-degree-of-freedom pose estimation method | |
| CN113720324A (en) | Octree map construction method and system | |
| Teixeira et al. | Real-time local 3D reconstruction for aerial inspection using superpixel expansion | |
| Mckenna et al. | Tracking human motion using auxiliary particle filters and iterated likelihood weighting | |
| CN110992424A (en) | Positioning method and system based on binocular vision | |
| CN110335308A (en) | The binocular vision speedometer calculation method examined based on disparity constraint and two-way annular | |
| Fan et al. | Dynamicfilter: an online dynamic objects removal framework for highly dynamic environments | |
| CN114494150A (en) | Design method of monocular vision odometer based on semi-direct method | |
| Willersinn et al. | Robust obstacle detection and tracking by motion analysis | |
| CN113063412B (en) | Multi-robot cooperative positioning and mapping method based on reliability analysis | |
| CN110864670B (en) | Method and system for acquiring position of target obstacle | |
| Cigla et al. | Gaussian mixture models for temporal depth fusion | |
| CN116129318A (en) | Unsupervised monocular three-dimensional target detection method based on video sequence and pre-training instance segmentation | |
| Zieliński et al. | 3d dense mapping with the graph of keyframe-based and view-dependent local maps | |
| CN113158816A (en) | Visual odometer quadric-surface road sign construction method for outdoor scene object | |
| Gandhi et al. | Motion analysis of omni-directional video streams for a mobile sentry | |
| Belter et al. | Keyframe-Based local normal distribution transform occupancy maps for environment mapping | |
| Salah et al. | Summarizing large scale 3D mesh for urban navigation | |
| CN115311424B (en) | Three-dimensional reconstruction method and device of target scene, unmanned aerial vehicle and storage medium |
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 |