CN109901207A - A kind of high-precision outdoor positioning method of Beidou satellite system and feature combinations - Google Patents
A kind of high-precision outdoor positioning method of Beidou satellite system and feature combinations Download PDFInfo
- Publication number
- CN109901207A CN109901207A CN201910198384.5A CN201910198384A CN109901207A CN 109901207 A CN109901207 A CN 109901207A CN 201910198384 A CN201910198384 A CN 201910198384A CN 109901207 A CN109901207 A CN 109901207A
- Authority
- CN
- China
- Prior art keywords
- image
- reference picture
- satellite system
- beidou satellite
- feature combinations
- 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
Landscapes
- Navigation (AREA)
Abstract
The invention belongs to technical field of navigation and positioning, disclose the high-precision outdoor positioning method of a kind of Beidou satellite system and feature combinations, comprising the following steps: acquisition reference picture；Reference picture is handled；According to treated, reference picture establishes database；Primary Location result is obtained using Beidou satellite system；Bit image undetermined is obtained, in conjunction with Primary Location as a result, the reference picture in bit image undetermined and data library is carried out similarity calculation, the reference picture to be matched is searched for, obtains vision auxiliary positioning result.The present invention, which solves the problems, such as that satellite-signal when intervisibility situation difference in the prior art is weak, causes positioning accuracy lower.
Description
Technical field
The present invention relates to technical field of navigation and positioning more particularly to the height of a kind of Beidou satellite system and feature combinations
Precision outdoor positioning method.
Background technique
Global Satellite Navigation System (guide number SS) is the great space and basis for IT application facility of country, in national economy
Construction and national defense safety field played an important role.After the GPS in the U.S., Russian GLONASS system and Europe
After Galileo system, China is just greatly developing Beidou satellite navigation system with independent intellectual property rights.Satellite navigation is from list
One system monopolization enters the new era of more GNSS competition and cooperation.Beidou satellite navigation system by space segment, ground segment and
User segment three parts composition, can round-the-clock in the world, round-the-clock be all types of user provide high-precision, it is highly reliable position,
Navigation, time service service, and have short message communication capacity, tentatively have area navigation, positioning and time service ability, positioning accuracy
Up to 10 meters.
Currently, the identification of vision place is an important issue for computer vision and robot.It is fixed with vision
Position is different, and the identification of vision place is mainly realized by image retrieval technologies and characteristic matching.Images match, which refers to, to be passed through
Certain matching algorithm identifies same place between two width or multiple image, as in two dimensional image matching by comparing target area and
The related coefficient of the window of same size in the field of search takes window center point conduct corresponding to related coefficient maximum in the field of search
Same place.Traditional image retrieval is mainly to pass through the method for extracting local feature and feature clustering coding to generate feature description
Son carries out image retrieval by calculating the distance of the Feature Descriptor between image.It is based on convolutional neural networks recent years
It is obtained a very large progress with the image search method of deep learning, realizes good effect on the problem of vision place identifies
Fruit.
Although Satellite Navigation Technique is quickly grown, in some built-up locations, intervisibility situation is poor, satellite navigation effect
Fruit is little.
Summary of the invention
The embodiment of the present application is by providing the high-precision outdoor positioning side of a kind of Beidou satellite system and feature combinations
Method, solving the problems, such as that satellite-signal when intervisibility situation difference in the prior art is weak causes positioning accuracy lower.
The embodiment of the present application provides the high-precision outdoor positioning method of a kind of Beidou satellite system and feature combinations, packet
Include following steps:
Step 1, acquisition reference picture；
Step 2 handles the reference picture；
Step 3, according to treated, the reference picture establishes database；
Step 4 obtains Primary Location result using Beidou satellite system；
Step 5 obtains bit image undetermined, in conjunction with the Primary Location as a result, by the bit image undetermined and the data
The reference picture in library carries out similarity calculation, searches for the reference picture to be matched, obtains vision auxiliary positioning knot
Fruit.
Preferably, the high-precision outdoor positioning method of the Beidou satellite system and feature combinations further include: step
6, determine in conjunction with the vision auxiliary positioning as a result, obtaining accurate auxiliary using SegNet image, semantic cutting techniques and PnP algorithm
Position result.
Preferably, the high-precision outdoor positioning method of the Beidou satellite system and feature combinations further include: step
7, real scene navigation is carried out according to the accurate auxiliary positioning result.
Preferably, in step 1, the location coordinate information of the reference picture is obtained by RTK technology, passes through IMU technology
Obtain the posture and angle information of the reference picture.
Preferably, in step 2, characteristics of image is carried out to the reference picture using NetVLAD image retrieval algorithm and is gathered
Class and coding obtain the Feature Descriptor of the reference picture, convert normalized vector for the reference picture.
Preferably, in step 3, database is established using the MIM indexing means based on R tree construction.
Preferably, in step 6, the implementation method of SegNet image, semantic cutting techniques is used are as follows: reduce described to be positioned
Image records the position of characteristic point weight limit, is calculated using the method for posterior probability each pixel, acquisition is calculated
Meet the image of maximum probability.
Preferably, in step 6, the implementation method of PnP algorithm is used are as follows: the bit image undetermined is resolved using PnP algorithm,
Simple multi-dimensional matrix is turned to, the attitude parameter minimized when re-projection error solves the image taking to be positioned is utilized.
Preferably, in step 7, after obtaining the accurate auxiliary positioning result, optimal path hair is calculated by server
Toward user terminal, path instruction is displayed on the screen by user terminal, and assists display neighboringly target POI information.
Preferably, the POI information includes the title, classification of point of interest, coordinate, detailed description in the setting range of position；
The point of interest includes famous scenery, landmark building, the scenic spots and historical sites.
One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:
In the case where intervisibility situation is poor, satellite positioning precision is poor, and in response to this, the present invention passes through benefit simultaneously
It is positioned with satellite information and ambient image information, not only to intervisibility situation is poor due to high building blocks instantly but characteristic point is abundant
City in position highly effective, and following be also of great significance to fields such as automatic Pilots.The present invention passes through elder generation
User's approximate location is determined with Beidou satellite navigation system BDS, the photo and the regional database figure then shot by user
As carrying out Image Feature Matching, so that it is determined that final position, the precision of satellite positioning is mentioned when this not only makes intervisibility situation poor
Height, and since satellite provides approximate location, the range of images match is reduced, to also accelerate the speed of images match
Degree.
Detailed description of the invention
It, below will be to needed in embodiment description in order to illustrate more clearly of the technical solution in the present embodiment
Attached drawing is briefly described, it should be apparent that, the accompanying drawings in the following description is one embodiment of the present of invention, general for this field
For logical technical staff, without creative efforts, it is also possible to obtain other drawings based on these drawings.
Fig. 1 is a kind of high-precision outdoor positioning of Beidou satellite system and feature combinations provided in an embodiment of the present invention
The overall framework schematic diagram of method；
Fig. 2 is a kind of high-precision outdoor positioning of Beidou satellite system and feature combinations provided in an embodiment of the present invention
IMU resolves the flow chart of position and attitude in method；
Fig. 3 is a kind of high-precision outdoor positioning of Beidou satellite system and feature combinations provided in an embodiment of the present invention
NetVLAD generates the block schematic illustration of image feature vector in method；
Fig. 4 is a kind of high-precision outdoor positioning of Beidou satellite system and feature combinations provided in an embodiment of the present invention
The block schematic illustration of the high accuracy positioning of view-based access control model in method；
Fig. 5 is a kind of high-precision outdoor positioning of Beidou satellite system and feature combinations provided in an embodiment of the present invention
Pass through the block schematic illustration of cell phone application application output navigation information in method.
Specific embodiment
In order to better understand the above technical scheme, in conjunction with appended figures and specific embodiments to upper
Technical solution is stated to be described in detail.
The high-precision outdoor positioning method for present embodiments providing a kind of Beidou satellite system and feature combinations, referring to
Fig. 1 is mainly comprised the steps that
(1) reference picture (i.e. database images) are acquired, is surveyed using real-time dynamic carrier phase difference RTK technology and inertia
It measures unit IMU (Inertial measurementunit) technology and obtains the exact position of image and the posture and angle of shooting.
Wherein, database diagram image position is determined using RTK technology high-precision, determine that database images are clapped using IMU technology
Posture and angle when taking the photograph.I.e. reference picture with accurate location information, and with specific direction and posture in addition to believing
Breath.
(2) database images are handled, and carry out the cluster and coding of characteristics of image to image using NetVLAD algorithm.
Wherein, the Feature Descriptor of database images is obtained using NetVLAD algorithm.
(3) database is established, image data is converted into multidimensional vector data, establishes database using MIM method.
(4) Primary Location carries out Primary Location using Beidou satellite system, due to defending merely when intervisibility situation is poor
Star positioning accuracy is not high, therefore, often obtains the approximate location of user at this time.
(5) vision auxiliary positioning, by the bit image undetermined image of shooting (i.e. user) using NetVLAD algorithm at
Reason, to obtain iamge description, the image in database established with step (2) carries out similarity calculation, finds and belongs to together
The picture of one example, to realize vision auxiliary positioning.
Wherein, the rough position and orientation that user shoots image are obtained by image recognition；Then it is shot using user
The geometrical relationship of image and database images carries out high accuracy positioning.
Since reference picture has specific posture information, the image ratio in the image and database of user's shooting
After completion, user current posture and direction can be obtained.When in face of some crossings for turning to complexity, it can solve well
The certainly critical problem that crossing turns to, and then the navigation Service of higher precision is provided.
(6) accurate auxiliary positioning, further increases positioning accurate using SegNet image, semantic cutting techniques and PnP algorithm
Degree, and prepare for subsequent real scene navigation.
(7) navigation information is exported, cell phone application is created, is positioned using abovementioned steps, is then handed over by data information
Mutually, it realizes and marks navigational arrows on real picture.
It is only labeled on two dimensional navigation electronic map relative to common navigation software, it is not easy to identify direction
Situation.The present invention is realized by data information interaction marks navigational arrows on real picture, obtain the current posture of user and
After, user can easily find correct path direction.The navigation information that the present invention exports has more intuitive, with user
Interactivity it is stronger.
I.e. the present invention first passes through big-dipper satellite and positions determining index range, and then user is transmitted to server after shooting picture
It holds, index compares image vector in database, attitude algorithm is carried out using SegNet image, semantic cutting techniques, PnP algorithm, from
And obtain the high precision position of user.
The present invention tentatively positions user using big-dipper satellite, determines locating approximate location, database index range is contracted
It is small, reduce the calculation amount that database carries out image feature vector retrieval.Compared in conjunction with the photo of existing photo and lane database
Pair as a result, to the resolving that the click-through line position to be positioned is set, by combining a variety of location informations, to obtain determining for higher precision
Position for real scene navigation as a result, lay the foundation.
Above-mentioned steps are described further below.
Step 1, image is acquired, exact position and the photograph angle of reference picture are obtained using RTK technology and IMU technology.
RTK reduces flowing using the spatial coherence of observation error between base station and rover station by way of difference
Most of error in observation data of standing, and then obtain high-precision coordinate.
Its basic double difference observation equation is writeable are as follows:
In formula,For double difference operator；λ is carrier wavelength；Double difference carrier wave phase between rover station and base station
Position observation；For corresponding double difference integer ambiguity parameter；Respectively flowing
Clean remaining orbit error, remaining ionosphere delay and remnants are not eliminated yet after seeking double difference between dynamic station and base station
Tropospheric delay, these errors are all related with the distance between rover station and base station；∑δiIt is then Multipath Errors, measurement is made an uproar
The sum of sound equal error.
The assembled unit that IMU is made of 3 accelerometers and 3 gyroscopes, obtains current image by accelerometer
Gravity direction, with magnetometer obtain picture horizontal azimuth, and using known map making be auxiliary information, to obtain figure
The posture and angle of piece.Since handheld device is based on local coordinate system, therefore converted using Quaternion method to global coordinate system,
Attitude algorithm is completed, process is as shown in Figure 2.
Quaternary number Q is expressed as follows:
Wherein: q0, q1, q2, q3For scalar,For vector.Quaternion differential equation can be obtained by derivation transformation,
It is as follows to be rewritten as matrix form:
Wherein: ωbFor angular speed.The differential equation can be solved with fourth-order Runge-Kutta method, have solution as follows:
The quaternary number at each moment can be calculated by above formula, and then it is as follows to can get corresponding attitude matrix:
Step 2, the image collected in step 1 is handled, using convolutional neural networks and VLAD algorithm to figure
Cluster and coding as carrying out characteristics of image.
NetVLAD is a kind of new image retrieval algorithm for combining convolutional neural networks and feature clustering coding, mistake
Journey are as follows: process of convolution first is carried out to image using trained convolutional neural networks model and obtains characteristic image, is recycled
VLAD algorithm carries out the cluster and coding of characteristics of image, it is made to meet the requirement of extensive place identification, detailed process such as Fig. 3 institute
Show.
The difference of all features and cluster centre is done cumulative, is extended to a normalized vector.Single compact spy
Sign vector is more easily implemented index, enhancing can discriminating power, raising search speed.Treatment process is as follows:
(1) it gives N number of D dimension topography and is described in xiAs input, K cluster centre ckAs the parameter of VLAD, VLAD
Output be a K × D dimension iamge description vector, be denoted as the matrix that V is a K × D, convert thereof into vector expression, and
It is normalized, calculation formula is as follows:
Wherein, xi(j) and ck(j) expression is j-th of characteristic value of i-th of local description and k-th of cluster centre.ak
(xi) it is the weight that i-th of local feature belongs to k-th of cluster.
(2) in traditional VLAD, due to ak(xi) it is a discontinuous value, value is 1 or 0, and is metSo that being unable to backpropagation.Using a kind of approximate mode, to ak(xi) it is soft distribution (soft
Assignment) such as following formula:
(3) a probability function weight is generated according to the distance of each local feature to cluster centre.For a part
Feature describes xiFor the range of weight under each clustering cluster between 0~1, weight highest can indicate this feature from clustering cluster
The cluster of the heart is nearest, and weight is low farther out from cluster center.Further obtain following formula:
Wherein, ωk′=2 α ck, bk=-α | | ck||2, finally, obtain VLAD feature vector are as follows:
Step 3, the image data after the completion of being handled using step 2, establishes database, and image data is converted to multidimensional arrow
Data are measured, are established using MIM indexing means (Multidimensional Indexing Method, multi-dimensional indexing method) method
Database.
Common location technology image coordinate acquired in the poor situation of intervisibility situation and attitude accuracy be not high, subsequent
Navigation information can have large error.In order to realize high-precision positioning and place identification, needing to establish one has image
Accurate geographic position coordinate and angular pose database, and data in database are examined using multi-dimensional indexing method
Rope, according to divide to data carry out cluster and using divide to search space carry out beta pruning to improve search efficiency, at present success
MIM method be the indexing means based on R tree construction.
The node structure of R tree can be described as follows:
Leaf node: (COUNT, LEVEL, < OI1, MBR1>,<OI2, MBR2>...,<OIm, MBRm>)。
Intermediate node: (COUNT, LEVEL, < CP1, MBR1>,<CP2, MBR2>...,<CPm, MBRm>)。
Wherein, < OIi, MBRi> be shelf space target data item, OIiWhat is saved is the characteristic information of the extraterrestrial target,
What MBRi was saved is the minimum area-encasing rectangle of the extraterrestrial target；<CPi, MBRi> be shelf space target index entry, CPiIt saves
The pointer for being directed to next stage subtree, MBRiWhat is saved is the minimum area-encasing rectangle i.e. minimum index sky of next stage subtree
Between.COUNT≤M indicates the number of the index entry stored or data item in R tree node, and LEVEL >=0 indicates the node in R tree
The number of plies.
Step 4, Primary Location is realized using Beidou satellite system.Since outdoor positioning environment is complicated, when circumstance of occlusion, has
Occur, such as the woods are dense, the towering intervisibility situation that easily leads to of building is poor, so that GPS and Big Dipper satellite signal are weak, satellite positioning essence
It spends not high.And the time interval of the position of satellite and navigation message arrival receiver is to solve for the key of location information, due to
Intervisibility situation is poor, necessarily leads to the time interval obtained inaccuracy, so when only obtain the rough range of user location.
Step 5, the image in the image and database of user terminal shooting is carried out similarity meter by visual information auxiliary positioning
It calculates, finds the picture for belonging to same instance, to realize that place identifies.This step needs to combine big-dipper satellite Primary Location result
It carries out.
In this step, according to the image data base established in step 2 characteristics of image coding obtained and step 3, then benefit
The Feature Descriptor of image is calculated with the NetVLAD algorithm based on convolutional neural networks and VLAD.It is indicated, will be used based on this image
Image in the image and database of the shooting of family end carries out similarity calculation, the picture for belonging to the same area is found, to pass through
Acquisition of vision information user current location.
In similarity calculation, using the distance between image feature vector is calculated, the same area is obtained to compare
Picture.A might as well be first set, b is the corresponding feature vector of two images, ai, biRespectively represent characteristic component.
Based on LpThe Minkowsky distance that norm defines are as follows:
If (i) p=1, become absolute value distance
(ii) if p=2, become Euclidean distance
By calculating above-mentioned distance, compares and obtain minimum value, can be obtained the image consistent with target area, further
Obtain its location information.
Step 6, accurate assisted location method is further increased using SegNet image, semantic cutting techniques and PnP algorithm
Positioning accuracy.
According to the location information tentatively obtained in step 5, we also use SegNet image, semantic cutting techniques and PnP
Method constrains position, keeps precision higher.SegNet is a kind of specifically with the image, semantic segmentation skill of deep learning
Art is encoded after image input, in an encoding process, is extracted feature by convolution, is kept image original size after convolution
Further progress compares.In this step, the scene information that SegNet image, semantic cutting techniques obtain, on the one hand can be used as
On the other hand the auxiliary information of positioning can also provide better user's interaction effect.
Further, we establish sparse cloud according to matched database image, are then calculated using sparse cloud and PnP
Method obtains exact position and the posture of user terminal image.Main point or less three steps carry out:
I. place identification obtains rough position and orientation；
Ii. high accuracy positioning is carried out using the relationship of mobile telephone image and database image；
Iii. unknown constraint is obtained using the semantic information of scenery.
Wherein, the present invention uses the SURF operator with rotation, scaling invariance to carry out feature extraction to benchmark image, from
And mobile telephone image is matched with the image in database.PnP derivation algorithm is commonly used in the posture tracking part of the front end SLAM
One of algorithm, not only calculation amount is small, can effectively shorten runing time, but also denoises better effect.
Step 5,6 flow chart see Fig. 4.
Step 7, the output method of navigation information creates cell phone application, is realized on real picture by data information interaction
Mark navigational arrows.
The present invention will be shown that POI shows function: mobile phone sensor is got mobile phone using POI module to information
The information such as position, direction, angle and both POI for being grabbed in Baidu map API comparison, while obtained image being shown
In the smart phone of client.And the data multiplicity shown, such as title, distance, size, while there are also corresponding directions
Arrow so that the image of POI show it is more outstanding.
Before realizing POI bandwagon effect, need to obtain the position of user and navigation information, then in coordinate
It is converted in system, obtains the virtual data information in coordinate system.It, can by the innovation and optimization of algorithm and the design at interface
To provide more smooth, comfortable experience for user.
The flow chart of this step is shown in Fig. 5.
When it is implemented, software mode, which can be used, in above step provides automatic running.
To sum up, the invention discloses a kind of a kind of outdoor positioning sides that place identification is realized by extracting Image Feature Matching
Method.The accurate geographic position and posture for obtaining reference picture by RTK technology and IMU technology first, then using based on convolution
The algorithm of neural network and VLAD carry out image procossing, and store in the database.The rough of user is positioned according to big-dipper satellite
Then image retrieval technologies, SegNet image, semantic cutting techniques and PnP algorithm auxiliary positioning are used, finally by data in position
The transmission of information exchange realization navigation information.The present invention is encoded by characteristics of image, and image is compared, so that place matches
The constraint of signal strength or weakness is got rid of, and user's posture and direction can determine whether solve a great problem of pedestrian's real scene navigation.This
The method that invention is combined using satellite navigation with Image Feature Matching provides high-precision outdoor positioning for user and navigation takes
Business.
The high-precision outdoor positioning method of a kind of Beidou satellite system provided in an embodiment of the present invention and feature combinations
Including at least following technical effect:
(1) present invention use the location navigation mode of much information fusion that can provide for user and more intuitive accurately guides
Service, it is not only highly effective to position in built-up city instantly, and also have to fields such as the following automatic Pilots
It is significant.
(2) then the present invention determines most final position by Image Feature Matching by first determining approximate location with big-dipper satellite
It sets, the precision of satellite Primary Location is not only made to be improved to some extent, but also accelerate the speed of images match.
It should be noted last that the above specific embodiment is only used to illustrate the technical scheme of the present invention and not to limit it,
Although being described the invention in detail referring to example, those skilled in the art should understand that, it can be to the present invention
Technical solution be modified or replaced equivalently, without departing from the spirit and scope of the technical solution of the present invention, should all cover
In the scope of the claims of the present invention.
Claims (10)
1. a kind of high-precision outdoor positioning method of Beidou satellite system and feature combinations, which is characterized in that including following
Step:
Step 1, acquisition reference picture；
Step 2 handles the reference picture；
Step 3, according to treated, the reference picture establishes database；
Step 4 obtains Primary Location result using Beidou satellite system；
Step 5 obtains bit image undetermined, in conjunction with the Primary Location as a result, by the bit image undetermined and the database
The reference picture carry out similarity calculation, search for the reference picture that is matched, obtain vision auxiliary positioning result.
2. the high-precision outdoor positioning method of Beidou satellite system according to claim 1 and feature combinations, special
Sign is, further includes:
Step 6, in conjunction with the vision auxiliary positioning as a result, SegNet image, semantic cutting techniques and PnP algorithm is used to obtain essence
True auxiliary positioning result.
3. the high-precision outdoor positioning method of Beidou satellite system according to claim 2 and feature combinations, special
Sign is, further includes:
Step 7 carries out real scene navigation according to the accurate auxiliary positioning result.
4. the high-precision outdoor positioning method of Beidou satellite system according to claim 1 and feature combinations, special
Sign is, in step 1, the location coordinate information of the reference picture is obtained by RTK technology, by described in the acquisition of IMU technology
The posture and angle information of reference picture.
5. the high-precision outdoor positioning method of Beidou satellite system according to claim 1 and feature combinations, special
Sign is, in step 2, carries out the cluster and coding of characteristics of image to the reference picture using NetVLAD image retrieval algorithm,
The reference picture is converted normalized vector by the Feature Descriptor for obtaining the reference picture.
6. the high-precision outdoor positioning method of Beidou satellite system according to claim 1 and feature combinations, special
Sign is, in step 3, establishes database using the MIM indexing means based on R tree construction.
7. the high-precision outdoor positioning method of Beidou satellite system according to claim 2 and feature combinations, special
Sign is, in step 6, uses the implementation method of SegNet image, semantic cutting techniques are as follows: reduces the bit image undetermined, record
The position of characteristic point weight limit calculates each pixel using the method for posterior probability, and acquisition, which calculates, meets probability
Maximum image.
8. the high-precision outdoor positioning method of Beidou satellite system according to claim 2 and feature combinations, special
Sign is, in step 6, uses the implementation method of PnP algorithm are as follows: resolves the bit image undetermined using PnP algorithm, turns to simple
Multi-dimensional matrix utilizes the attitude parameter minimized when re-projection error solves the image taking to be positioned.
9. the high-precision outdoor positioning method of Beidou satellite system according to claim 3 and feature combinations, special
Sign is, in step 7, after obtaining the accurate auxiliary positioning result, calculates optimal path by server and is sent to user terminal,
Path instruction is displayed on the screen by user terminal, and assists display neighboringly target POI information.
10. the high-precision outdoor positioning method of Beidou satellite system according to claim 9 and feature combinations, special
Sign is that the POI information includes the title, classification of point of interest, coordinate, detailed description in the setting range of position；The interest
Point includes famous scenery, landmark building, the scenic spots and historical sites.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910198384.5A CN109901207A (en) | 2019-03-15 | 2019-03-15 | A kind of high-precision outdoor positioning method of Beidou satellite system and feature combinations |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910198384.5A CN109901207A (en) | 2019-03-15 | 2019-03-15 | A kind of high-precision outdoor positioning method of Beidou satellite system and feature combinations |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109901207A true CN109901207A (en) | 2019-06-18 |
Family
ID=66953340
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910198384.5A Pending CN109901207A (en) | 2019-03-15 | 2019-03-15 | A kind of high-precision outdoor positioning method of Beidou satellite system and feature combinations |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109901207A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110509273A (en) * | 2019-08-16 | 2019-11-29 | 天津职业技术师范大学(中国职业培训指导教师进修中心) | The robot mechanical arm of view-based access control model deep learning feature detects and grasping means |
| CN110658809A (en) * | 2019-08-15 | 2020-01-07 | 北京致行慕远科技有限公司 | Method and device for processing travelling of movable equipment and storage medium |
| CN111025364A (en) * | 2019-12-17 | 2020-04-17 | 南京航空航天大学 | Machine vision positioning system and method based on satellite assistance |
| CN111103612A (en) * | 2019-12-27 | 2020-05-05 | 北京车和家信息技术有限公司 | Visual self-positioning method and device and vehicle |
| CN111141295A (en) * | 2019-12-20 | 2020-05-12 | 南京航空航天大学 | Automatic map recovery method based on monocular ORB-SLAM |
| CN111696311A (en) * | 2020-06-23 | 2020-09-22 | 广东小天才科技有限公司 | Indoor and outdoor auxiliary identification method, intelligent wearable device and storage medium |
| CN113739797A (en) * | 2020-05-31 | 2021-12-03 | 华为技术有限公司 | Visual positioning method and device |
| CN114199275A (en) * | 2020-09-18 | 2022-03-18 | 阿里巴巴集团控股有限公司 | Parameter determination method and device for sensor |
| CN115293301A (en) * | 2022-10-09 | 2022-11-04 | 腾讯科技（深圳）有限公司 | Estimation method and device for lane change direction of vehicle and storage medium |
| CN117092678A (en) * | 2023-10-18 | 2023-11-21 | 深圳市硕腾科技有限公司 | High-precision RTK positioning system based on LoRa |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107084727A (en) * | 2017-04-12 | 2017-08-22 | 武汉理工大学 | A kind of vision positioning system and method based on high-precision three-dimensional map |
| CN107704821A (en) * | 2017-09-29 | 2018-02-16 | 河北工业大学 | A kind of vehicle pose computational methods of bend |
| CN107742311A (en) * | 2017-09-29 | 2018-02-27 | 北京易达图灵科技有限公司 | A kind of method and device of vision positioning |
| CN108765326A (en) * | 2018-05-18 | 2018-11-06 | 南京大学 | A kind of synchronous superposition method and device |
| CN108802785A (en) * | 2018-08-24 | 2018-11-13 | 清华大学 | Vehicle method for self-locating based on High-precision Vector map and monocular vision sensor |
| CN109255381A (en) * | 2018-09-06 | 2019-01-22 | 华南理工大学 | A kind of image classification method based on the sparse adaptive depth network of second order VLAD |
-
2019
- 2019-03-15 CN CN201910198384.5A patent/CN109901207A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107084727A (en) * | 2017-04-12 | 2017-08-22 | 武汉理工大学 | A kind of vision positioning system and method based on high-precision three-dimensional map |
| CN107704821A (en) * | 2017-09-29 | 2018-02-16 | 河北工业大学 | A kind of vehicle pose computational methods of bend |
| CN107742311A (en) * | 2017-09-29 | 2018-02-27 | 北京易达图灵科技有限公司 | A kind of method and device of vision positioning |
| CN108765326A (en) * | 2018-05-18 | 2018-11-06 | 南京大学 | A kind of synchronous superposition method and device |
| CN108802785A (en) * | 2018-08-24 | 2018-11-13 | 清华大学 | Vehicle method for self-locating based on High-precision Vector map and monocular vision sensor |
| CN109255381A (en) * | 2018-09-06 | 2019-01-22 | 华南理工大学 | A kind of image classification method based on the sparse adaptive depth network of second order VLAD |
Non-Patent Citations (3)
| Title |
|---|
| LIFENG AN ETC.: ""Semantic segmentation-aided visual odometry for urban autonomous driving"", 《INTERNATIONAL JOURNAL OF ADVANCED ROBOTIC SYSTEMS》 * |
| 卞贤掌等: ""基于语义分割的增强现实图像配准技术"", 《电子技术与软件工程》 * |
| 王浩: ""基于图像检索的定位算法研究与实现"", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110658809A (en) * | 2019-08-15 | 2020-01-07 | 北京致行慕远科技有限公司 | Method and device for processing travelling of movable equipment and storage medium |
| CN110509273B (en) * | 2019-08-16 | 2022-05-06 | 天津职业技术师范大学(中国职业培训指导教师进修中心) | Robot manipulator detection and grabbing method based on visual deep learning features |
| CN110509273A (en) * | 2019-08-16 | 2019-11-29 | 天津职业技术师范大学(中国职业培训指导教师进修中心) | The robot mechanical arm of view-based access control model deep learning feature detects and grasping means |
| CN111025364A (en) * | 2019-12-17 | 2020-04-17 | 南京航空航天大学 | Machine vision positioning system and method based on satellite assistance |
| CN111025364B (en) * | 2019-12-17 | 2023-05-16 | 南京航空航天大学 | Machine vision positioning system and method based on satellite assistance |
| CN111141295A (en) * | 2019-12-20 | 2020-05-12 | 南京航空航天大学 | Automatic map recovery method based on monocular ORB-SLAM |
| CN111103612A (en) * | 2019-12-27 | 2020-05-05 | 北京车和家信息技术有限公司 | Visual self-positioning method and device and vehicle |
| CN113739797A (en) * | 2020-05-31 | 2021-12-03 | 华为技术有限公司 | Visual positioning method and device |
| CN111696311A (en) * | 2020-06-23 | 2020-09-22 | 广东小天才科技有限公司 | Indoor and outdoor auxiliary identification method, intelligent wearable device and storage medium |
| CN114199275A (en) * | 2020-09-18 | 2022-03-18 | 阿里巴巴集团控股有限公司 | Parameter determination method and device for sensor |
| CN115293301A (en) * | 2022-10-09 | 2022-11-04 | 腾讯科技（深圳）有限公司 | Estimation method and device for lane change direction of vehicle and storage medium |
| CN117092678A (en) * | 2023-10-18 | 2023-11-21 | 深圳市硕腾科技有限公司 | High-precision RTK positioning system based on LoRa |
| CN117092678B (en) * | 2023-10-18 | 2024-01-26 | 深圳市硕腾科技有限公司 | High-precision RTK positioning system based on LoRa |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109901207A (en) | A kind of high-precision outdoor positioning method of Beidou satellite system and feature combinations | |
| CN110322511A (en) | A kind of semantic SLAM method and system based on object and plane characteristic | |
| CN109658445A (en) | Network training method, increment build drawing method, localization method, device and equipment | |
| CN104422439B (en) | Air navigation aid, device, server, navigation system and its application method | |
| CN104200523B (en) | A kind of large scene three-dimensional rebuilding method for merging additional information | |
| Brejcha et al. | State-of-the-art in visual geo-localization | |
| CN103632626A (en) | Intelligent tour guide realizing method and intelligent tour guide device based on mobile network and mobile client | |
| CN112445929B (en) | Visual positioning method and related device | |
| CN111256696B (en) | Aircraft autonomous navigation method with multi-feature and multi-level scene matching | |
| Vishal et al. | Accurate localization by fusing images and GPS signals | |
| US11915478B2 (en) | Bayesian methodology for geospatial object/characteristic detection | |
| Feng et al. | Visual Map Construction Using RGB‐D Sensors for Image‐Based Localization in Indoor Environments | |
| CN113239952A (en) | Aerial image geographical positioning method based on spatial scale attention mechanism and vector map | |
| CN116839600A (en) | Visual mapping navigation positioning method based on lightweight point cloud map | |
| Wilson et al. | Visual and object geo-localization: A comprehensive survey | |
| CN108534781A (en) | Indoor orientation method based on video | |
| Sui et al. | An accurate indoor localization approach using cellphone camera | |
| CN109902139B (en) | R-tree-based track data compression method | |
| Li et al. | Vision-based indoor localization via a visual SLAM approach | |
| Yin et al. | A PCLR-GIST algorithm for fast image retrieval in visual indoor localization system | |
| Liu et al. | An Outdoor Pedestrian Localization Scheme Fusing PDR and VPR | |
| WO2023246537A1 (en) | Navigation method, visual positioning method, navigation map construction method, and electronic device | |
| CN115641499B (en) | Photographing real-time positioning method, device and storage medium based on street view feature library | |
| Tutzauer et al. | Processing of crawled urban imagery for building use classification | |
| CN117170501B (en) | Visual tracking method based on point-line fusion characteristics |
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: 20190618 |
|
| RJ01 | Rejection of invention patent application after publication |