CN110135387A - A kind of image rapid identification method based on sensor fusion - Google Patents
A kind of image rapid identification method based on sensor fusion Download PDFInfo
- Publication number
- CN110135387A CN110135387A CN201910437779.6A CN201910437779A CN110135387A CN 110135387 A CN110135387 A CN 110135387A CN 201910437779 A CN201910437779 A CN 201910437779A CN 110135387 A CN110135387 A CN 110135387A
- Authority
- CN
- China
- Prior art keywords
- coordinate system
- traffic sign
- axis
- coordinate
- sensor fusion
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
- G01C21/34—Route searching; Route guidance
- G01C21/3453—Special cost functions, i.e. other than distance or default speed limit of road segments
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/251—Fusion techniques of input or preprocessed data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/582—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of traffic signs
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Automation & Control Theory (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Multimedia (AREA)
- Traffic Control Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- Navigation (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of image rapid identification methods based on sensor fusion, this method obtains distance and bearing information of the automatic driving vehicle relative to traffic sign based on high-precision map off-line data and inertial navigation set location information in real time, and determine therefrom that the corresponding position of traffic sign in the picture, assist visual identification algorithm.The present invention assists vision positioning using high-precision map and location information, and target position prior information can be provided for visual identity, reduces calculation amount, improves accuracy of identification, and without increasing any hardware cost.
Description
Technical field
The present invention relates to integrated navigation and Traffic Sign Recognition technical field more particularly to it is a kind of based on sensor fusion
Image rapid identification method.
Background technique
Three intelligence have been held in 2004,2005 and 2007 as DARPA restarts unmanned vehicle technology research, and successively
Energy vehicle challenge match, Google put into automatic Pilot technical research energetically, have started automatic Pilot technology in global range in recent years and ground
The upsurge studied carefully, automatic Pilot field venture company also comes into being like the mushrooms after rain.
Automatic Pilot technology develops by more than ten years, gradually from laboratory research, to the contest of enclosed environment technology, again to solid
Determine scene Demonstration Application, although achieving considerable progress, still has relatively large distance apart from practical application, it can only be in closing scene
Simple certainty task is executed under state of driving at low speed, and reliability is lower.And it is general with urbanization progress and automobile
And urban traffic congestion aggravation, the unmanned traffic safety problem being likely to result in become to become increasingly conspicuous.
Driver assistance system based on computer vision is one of the important measures for solving traffic safety problem, is mainly existed
The fields such as Traffic Sign Recognition, road Identification play a significant role.Currently, Traffic Sign Recognition mainly uses computer vision
Real-time detection of the system under no any prior information, however since traffic sign is many kinds of, it is grabbed in the case of real-time detection
The picture background taken is complicated, and the influence of intensity of illumination and angle is likely to influence the correct Classification and Identification of traffic sign；Furthermore
Real-time detection also carries out identification work in no traffic sign, wastes a large amount of computing resource.
Summary of the invention
In view of the above deficiencies, provided herein is a kind of image rapid identification method based on sensor fusion, inertial navigation/height is utilized
Accuracy Figure digital map navigation obtains position of the traffic sign in picture in advance, can relatively accurately extract traffic sign key
Feature solves the problems, such as that there is no traffic sign prior information and computing resource wastes in conventional method.It is obtained in advance
After road signs information, machine learning or deep learning algorithm are recycled, Classification and Identification is carried out to traffic sign.
The technical solution adopted in the present invention is as follows: a kind of image rapid identification method based on sensor fusion, including
Following steps:
(1) according to inertial navigation and high-precision map, longitude, the latitude height, elevation information of vehicle-mounted inertial navigation and traffic sign are obtained,
By attitude matrix conversion and coordinate transform, relative position and the posture relationship of camera and traffic sign are obtained；
(2) according to the relationship of camera coordinates system and photo coordinate system, traffic sign is obtained in the position of photo coordinate system
It sets, and is converted into pixel coordinate system coordinate；
(3) according to traffic sign in pixel coordinate system position, extract picture in traffic sign important feature；
(4) traffic sign is identified.
Further, the step (1) is specific as follows:
(1.1) by inertial navigation online resolution, the real-time accuracy B of carrier is obtaineda, dimension La, height Ha, obtain accordingly high-precision
Spend corresponding traffic sign and its longitude B in maps, latitude Ls, height Hs, and by carrier and the warp of corresponding traffic sign
Degree, latitude, elevation information are respectively converted into the coordinate (x of WGS84 geocentric coordinate systemia,yia,zia) and (xis,yis,zis)；
(1.2) it establishes using inertial navigation carrier center as origin O, using east orientation as X-axis, north orientation is Y-axis, and day is to the navigation for Z axis
Coordinate system Oan-XanYanZan, converted by coordinate system, obtain traffic sign location information in navigational coordinate system Oan-XanYanZanIn
Coordinate (xan,yan,zan)；
(1.3) posture information exported according to inertial navigation: yaw angle φab, pitching angle thetaabWith roll angle γab, establish carrier appearance
State matrix, and by carrier navigational coordinate system O in step (1.2)an-XanYanZanIt is converted to carrier body coordinate system Oab-
XabYabZab, wherein center sensor is origin O, and preceding longitudinally axially is Y-axis, and being directed toward right side perpendicular to the longitudinal axis is X-axis, perpendicular to
O-XY plane is Z axis upwards, is converted to seat of the traffic sign location information in carrier body coordinate system by attitude matrix
Mark (xab,yab,zab)；
(1.4) it establishes using camera photocentre as origin, optical axis is the camera coordinates system O of Z axiscb-XcbYcbZcb, turned by posture
It changes matrix and obtains traffic sign location information in camera coordinates system Ocb-XcbYcbZcbCoordinate.
Further, the step (2) is specific as follows:
(2.1) according to camera coordinates system Ocb-XcbYcbZcbPlane of delineation coordinate system O-XY transformational relation and phase focal length f,
Traffic sign is obtained in plane of delineation coordinate system O-XY coordinate (x, y)；
(2.2) according to plane of delineation coordinate system O-XY- pixel coordinate system Ouv- UV relationship utilizes transition matrix and camera solution
Image force Rex、ReyTraffic sign is obtained in pixel coordinate system coordinate (u, v).
Further, the step (3) is specific as follows:
(3.1) centered on the traffic sign determined by step (2) is in pixel coordinate system position, the ruler of given range is intercepted
It is very little, accurately to obtain traffic sign feature.
Further, the step (4) is specific as follows:
(4.1) the training set original image size of acquisition is normalized using bicubic interpolation algorithm or other algorithms；
(4.2) contrast is enhanced to the picture after normalization, reducing intensity of illumination bring influences；
(4.3) treated picture and label are stored as TFRecord format；
(4.4) picture is identified using deep learning algorithm.
Further, the step (4.4) is specific as follows:
(4.4.1) is trained using the TFRecords data that deep learning model stores step (4.3)；
(4.4.2) computation model loss function, the loss function use cross entropy loss function；
(4.4.3) uses gradient descent method combination sliding average, is iterated calculating by objective function of loss function；
Trained model is stored as ckpt model by (4.4.4), is moved on unmanned vehicle to spy acquired in step (3)
Sign picture is identified.
Further, the deep learning model in the step (4.4.1) preferably be selected from AlexNet, SqueezeNet,
GooLeNet, ImageNet, Faster R-CNN, R-FCN, YOLO convolutional neural networks model.
Beneficial effects of the present invention are as follows: in automatic driving vehicle sensory perceptual system, the Traffic Sign Images of view-based access control model are known
It is not a key technology, still, since the limitation of current hardware computing resource and extraneous illumination condition change to picture quality
It influences, under the premise of no priori knowledge auxiliary, it is difficult to which that realizes traffic sign accurately identifies and be engineered application.For this
Problem, the invention proposes a kind of image rapid identification methods based on sensor fusion.By inertial navigation, satellite navigation, view
Feel, accurately diagram data fusion, to Traffic Sign Images information pre-processing, obtains target area, pass through deep learning model
Phase processor is realized and is identified to road signs information, is not being improved on calculation amount basis, is being greatly improved accuracy of identification, meanwhile, by
In getting target area by other sensors, influence of the light condition to image recognition, therefore, this hair can be effectively reduced
Bright model robustness is good, strong anti-interference performance, and advantage is more obvious in light condition complex environment.
Detailed description of the invention
Fig. 1 is camera coordinates system-image coordinate system relationship；
Fig. 2 is image coordinate system-pixel coordinate system relationship；
Fig. 3 is AlexNet network structure.
Specific embodiment
Below by embodiment, the present invention is described in further detail.
The present invention provides a kind of image rapid identification method based on sensor fusion, includes the following steps:
(1) inertial navigation/high-precision map supplementary guiding information resolves:
According to inertial navigation and high-precision map, longitude, the latitude height, elevation information of vehicle-mounted inertial navigation and traffic sign are obtained, is led to
Attitude matrix conversion and coordinate transform are crossed, relative position and the posture relationship of camera and traffic sign are obtained.When traffic sign with
The distance of camera and when meeting visual range condition with optical axis angle, camera is started to work.
(2) according to the relationship of camera coordinates system and photo coordinate system, traffic sign is obtained in the position of photo coordinate system
It sets, and is converted into pixel coordinate system coordinate；
(3) training dataset pre-processes: carrying out a series of processing to training dataset, makes its standardization, normalization.
(4) use machine learning or deep learning model (present invention is by taking AlexNet as an example) to step (3) treated hand over
Logical sign image is trained；
(5) traffic sign is obtained centered on pixel coordinate system position according to step (2), picture is intercepted as 224*224
Size is input to step (4) trained model, for identifying traffic sign classification.
Further, the step (1) is specific as follows:
(1.1) longitude and latitude height and WGS84 coordinate system are converted
Assuming that Ha、HsRespectively represent the height of inertial navigation carrier and traffic sign；Ba、BsRespectively represent inertial navigation carrier and traffic
The longitude of mark；La、LsThe latitude of inertial navigation carrier and traffic sign is respectively represented, then inertial navigation carrier and traffic sign be in WGS84
Coordinate under heart coordinate system is respectively as follows:
WhereinFor WGS84 earth ellipticity；a,b
Respectively WGS84 coordinate system earth long axis and short axle.
(1.2) WGS84 geocentric coordinate system and inertial navigation navigational coordinate system are converted
Assuming that Oi-XiYiZiFor WGS84 geocentric coordinate system, Oan-XanYanZanFor using inertial navigation receiving antenna as the navigation of dot
Coordinate system (northeast day), Oi-XiYiZiTo Oan-XanYanZanAttitude matrixAre as follows:
Wherein λ andIt is expressed as follows: geographic coordinate system Oi-XiYiZiRotation λ+90 is spent counterclockwise about the z axis, further around new coordinate system
X-axis rotates counterclockwiseObtain navigational coordinate system Oan-XanYanZan。
Therefore coordinate of the traffic sign under inertial navigation navigational coordinate system are as follows:
(1.3) inertial navigation navigational coordinate system and inertial navigation body coordinate system are converted
Assuming that Oab-XabYabZabFor inertial navigation body coordinate system, inertial navigation long axis is Y-axis, and short axle is X-axis, and origin and inertial navigation are navigated
Coordinate system Oan-XanYanZanIt is overlapped.Oan-XanYanZanTo Oab-XabYabZabAttitude matrixAre as follows:
Wherein φab、θab、γabRespectively inertial navigation yaw angle, pitch angle and roll angle, are exported by inertial navigation system.
Therefore coordinate of the traffic sign under antenna body coordinate system is acquired are as follows:
(1.4) inertial navigation navigational coordinate system and antenna body coordinate system are converted
To obtain traffic sign from inertial navigation body coordinate system Oab-XabYabZabTo camera body coordinate system Ocb-XcbYcbZcb's
Coordinate transformation relation needs to obtain Ocb-XcbYcbZcbWith Oab-XabYabZabThree attitude angles and three translational movements.Assuming that:
Wherein φcb、θcb、γcb、Δxcb、Δycb、ΔzcbFor Camera extrinsic.Therefore Ocb-XcbYcbZcbTo Oab-XabYabZab
Pose transformation matrixAre as follows:
Therefore coordinate of the traffic sign under inertial navigation body coordinate system is acquired are as follows:
Further, the step (2) is specific as follows:
Fig. 2 is camera coordinates system to the projection relation schematic diagram of photo coordinate system, and wherein p (x, y) is traffic mark
Will P (xcb,ycb,zcb) in the projection of photo coordinate system O-XY, f is focal length.According to Similar Principle of Triangle, can obtain:
Therefore it can be solved in the position of photo coordinate system in the position of camera coordinates system according to traffic sign by formula (9)
It sets.
Fig. 3 is the transformational relation schematic diagram of photo coordinate system and pixel coordinate system.Photo coordinate system origin O is in pixel
The position of coordinate system is (u0,v0), it is assumed that the resolving power of camera both horizontally and vertically is respectively RexAnd Rey, therefore ask
Obtain position of the traffic sign in pixel coordinate system are as follows:
Formula (9) mid-focal length f and u in formula (10)0、v0、Rex、ReyIt is camera internal reference.
Further, the step (4) is specific as follows:
(4.1) original image size is normalized to by 224*224 using bicubic interpolation algorithm, convenient for the reading of network；
(4.2) contrast is enhanced to the picture after normalization, reducing intensity of illumination bring influences；
(4.3) treated picture and label are stored as TFRecord format, to reduce memory space, acceleration model
Read-write and training speed.
(4.4) picture is identified using deep learning algorithm.
Further, the step (4.4) is specific as follows:
(4.4.1) using deep learning model (deep learning model can using from AlexNet, SqueezeNet,
GooLeNet, ImageNet, Faster R-CNN, R-FCN, YOLO convolutional neural networks model etc., use in the present embodiment
AlexNet convolutional neural networks model is as example) the TFRecords data that are stored to step (3) calculate；
(4.4.2) computation model loss function, the loss function use cross entropy loss function, are defined as follows:
(4.4.3) uses gradient descent method combination sliding average, is iterated calculating by objective function of loss function；
(4.4.4) will be trained, and model is stored as ckpt model, moves on unmanned vehicle to acquired in step (3)
Feature image is identified, persistence and transfer learning are convenient for.
The above description is only a preferred embodiment of the present invention, is not intended to limit the scope of the invention, all utilizations
Equivalent structure or equivalent flow shift made by description of the invention and accompanying drawing content is applied directly or indirectly in other correlations
Technical field, be included within the scope of the present invention.
Claims (7)
1. a kind of image rapid identification method based on sensor fusion, which comprises the steps of:
(1) according to inertial navigation and high-precision map, longitude, the latitude height, elevation information of vehicle-mounted inertial navigation and traffic sign is obtained, is passed through
Attitude matrix conversion and coordinate transform, obtain relative position and the posture relationship of camera and traffic sign；
(2) according to the relationship of camera coordinates system and photo coordinate system, traffic sign is obtained in the position of photo coordinate system, and
It is converted into pixel coordinate system coordinate；
(3) according to traffic sign in pixel coordinate system position, extract picture in traffic sign important feature；
(4) traffic sign is identified.
2. a kind of image rapid identification method based on sensor fusion according to claim 1, which is characterized in that described
Step (1) is specific as follows:
(1.1) by inertial navigation online resolution, the real-time accuracy B of carrier is obtaineda, dimension La, height Ha, obtain accordingly accurately
Corresponding traffic sign and its longitude B in figures, latitude Ls, height Hs, and by the longitude of carrier and corresponding traffic sign, latitude
Degree, elevation information are respectively converted into the coordinate (x of WGS84 geocentric coordinate systemia,yia,zia) and (xis,yis,zis)；
(1.2) it establishes using inertial navigation carrier center as origin O, using east orientation as X-axis, north orientation is Y-axis, and day is to the navigation coordinate for Z axis
It is Oan-XanYanZan, converted by coordinate system, obtain traffic sign location information in navigational coordinate system Oan-XanYanZanIn seat
Mark (xan,yan,zan)；
(1.3) posture information exported according to inertial navigation: yaw angle φab, pitching angle thetaabWith roll angle γab, establish attitude of carrier square
Battle array, and by carrier navigational coordinate system O in step (1.2)an-XanYanZanIt is converted to carrier body coordinate system Oab-XabYabZab,
Wherein center sensor is origin O, and preceding longitudinally axially is Y-axis, and being directed toward right side perpendicular to the longitudinal axis is X-axis, perpendicular to O-XY plane
It is upwards Z axis, coordinate (x of the traffic sign location information in carrier body coordinate system is converted to by attitude matrixab,
yab,zab)；
(1.4) it establishes using camera photocentre as origin, optical axis is the camera coordinates system O of Z axiscb-XcbYcbZcb, square is converted by posture
Battle array obtains traffic sign location information in camera coordinates system Ocb-XcbYcbZcbCoordinate.
3. a kind of image rapid identification method based on sensor fusion according to claim 1, which is characterized in that described
Step (2) is specific as follows:
(2.1) according to camera coordinates system Ocb-XcbYcbZcbPlane of delineation coordinate system O-XY transformational relation and phase focal length f are obtained
Traffic sign is in plane of delineation coordinate system O-XY coordinate (x, y)；
(2.2) according to plane of delineation coordinate system O-XY- pixel coordinate system Ouv- UV relationship utilizes transition matrix and camera resolving power
Rex、ReyTraffic sign is obtained in pixel coordinate system coordinate (u, v).
4. a kind of image rapid identification method based on sensor fusion according to claim 1, which is characterized in that described
Step (3) is specific as follows:
(3.1) centered on the traffic sign determined by step (2) is in pixel coordinate system position, the size of given range is intercepted,
Accurately to obtain traffic sign feature.
5. a kind of image rapid identification method based on sensor fusion according to claim 1, which is characterized in that described
Step (4) is specific as follows:
(4.1) the training set original image size of acquisition is normalized using bicubic interpolation algorithm or other algorithms；
(4.2) contrast is enhanced to the picture after normalization, reducing intensity of illumination bring influences；
(4.3) treated picture and label are stored as TFRecord format；
(4.4) picture is identified using deep learning algorithm.
6. a kind of image rapid identification method based on sensor fusion according to claim 5, which is characterized in that described
Step (4.4) is specific as follows:
(4.4.1) is trained using the TFRecords data that deep learning model stores step (4.3)；
(4.4.2) computation model loss function, the loss function use cross entropy loss function；
(4.4.3) uses gradient descent method combination sliding average, is iterated calculating by objective function of loss function；
Trained model is stored as ckpt model by (4.4.4), is moved on unmanned vehicle to characteristic pattern acquired in step (3)
Piece is identified.
7. a kind of image rapid identification method based on sensor fusion according to claim 6, which is characterized in that described
Deep learning model in step (4.4.1) preferably is selected from AlexNet, SqueezeNet, GooLeNet, ImageNet, Faster
R-CNN, R-FCN, YOLO convolutional neural networks model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910437779.6A CN110135387B (en) | 2019-05-24 | 2019-05-24 | Image rapid identification method based on sensor fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910437779.6A CN110135387B (en) | 2019-05-24 | 2019-05-24 | Image rapid identification method based on sensor fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110135387A true CN110135387A (en) | 2019-08-16 |
| CN110135387B CN110135387B (en) | 2021-03-02 |
Family
ID=67572988
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910437779.6A Active CN110135387B (en) | 2019-05-24 | 2019-05-24 | Image rapid identification method based on sensor fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110135387B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113465608A (en) * | 2021-07-22 | 2021-10-01 | 清华大学苏州汽车研究院（吴江） | Calibration method and system for roadside sensor |
| CN115164912A (en) * | 2022-06-24 | 2022-10-11 | 宁波均胜智能汽车技术研究院有限公司 | Vehicle position positioning method and device and readable storage medium |
| CN113465608B (en) * | 2021-07-22 | 2024-05-03 | 清华大学苏州汽车研究院（吴江） | Road side sensor calibration method and system |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103207090A (en) * | 2013-04-09 | 2013-07-17 | 北京理工大学 | Driverless vehicle environment simulation test system and test method |
| CN103942546A (en) * | 2014-05-08 | 2014-07-23 | 奇瑞汽车股份有限公司 | Guide traffic marking identification system and method in municipal environment |
| CN106326858A (en) * | 2016-08-23 | 2017-01-11 | 北京航空航天大学 | Road traffic sign automatic identification and management system based on deep learning |
| CN106407947A (en) * | 2016-09-29 | 2017-02-15 | 百度在线网络技术（北京）有限公司 | Target object recognition method and device applied to unmanned vehicle |
| CN106559755A (en) * | 2016-11-08 | 2017-04-05 | 长安大学 | The car networking electronic communication mark broadcasting method ensured based on QoS under collision detection |
| CN106560835A (en) * | 2015-09-30 | 2017-04-12 | 高德软件有限公司 | Guideboard identification method and device |
| CN107422730A (en) * | 2017-06-09 | 2017-12-01 | 武汉市众向科技有限公司 | The AGV transportation systems of view-based access control model guiding and its driving control method |
| CN107782335A (en) * | 2016-08-31 | 2018-03-09 | 北京自动化控制设备研究所 | A kind of contactless line detection system inertial navigation and laser range finder self-calibrating method |
| CN107957266A (en) * | 2017-11-16 | 2018-04-24 | 北京小米移动软件有限公司 | Localization method, device and storage medium |
| CN108628324A (en) * | 2018-07-12 | 2018-10-09 | 中国科学院深圳先进技术研究院 | Unmanned vehicle navigation method, device, equipment based on map vector and storage medium |
| CN108717712A (en) * | 2018-05-29 | 2018-10-30 | 东北大学 | A kind of vision inertial navigation SLAM methods assumed based on ground level |
| CN108983219A (en) * | 2018-08-17 | 2018-12-11 | 北京航空航天大学 | A kind of image information of traffic scene and the fusion method and system of radar information |
| CN109753949A (en) * | 2019-01-31 | 2019-05-14 | 南京邮电大学 | A kind of more form method for traffic sign detection based on deep learning |
-
2019
- 2019-05-24 CN CN201910437779.6A patent/CN110135387B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103207090A (en) * | 2013-04-09 | 2013-07-17 | 北京理工大学 | Driverless vehicle environment simulation test system and test method |
| CN103942546A (en) * | 2014-05-08 | 2014-07-23 | 奇瑞汽车股份有限公司 | Guide traffic marking identification system and method in municipal environment |
| CN106560835A (en) * | 2015-09-30 | 2017-04-12 | 高德软件有限公司 | Guideboard identification method and device |
| CN106326858A (en) * | 2016-08-23 | 2017-01-11 | 北京航空航天大学 | Road traffic sign automatic identification and management system based on deep learning |
| CN107782335A (en) * | 2016-08-31 | 2018-03-09 | 北京自动化控制设备研究所 | A kind of contactless line detection system inertial navigation and laser range finder self-calibrating method |
| CN106407947A (en) * | 2016-09-29 | 2017-02-15 | 百度在线网络技术（北京）有限公司 | Target object recognition method and device applied to unmanned vehicle |
| CN106559755A (en) * | 2016-11-08 | 2017-04-05 | 长安大学 | The car networking electronic communication mark broadcasting method ensured based on QoS under collision detection |
| CN107422730A (en) * | 2017-06-09 | 2017-12-01 | 武汉市众向科技有限公司 | The AGV transportation systems of view-based access control model guiding and its driving control method |
| CN107957266A (en) * | 2017-11-16 | 2018-04-24 | 北京小米移动软件有限公司 | Localization method, device and storage medium |
| CN108717712A (en) * | 2018-05-29 | 2018-10-30 | 东北大学 | A kind of vision inertial navigation SLAM methods assumed based on ground level |
| CN108628324A (en) * | 2018-07-12 | 2018-10-09 | 中国科学院深圳先进技术研究院 | Unmanned vehicle navigation method, device, equipment based on map vector and storage medium |
| CN108983219A (en) * | 2018-08-17 | 2018-12-11 | 北京航空航天大学 | A kind of image information of traffic scene and the fusion method and system of radar information |
| CN109753949A (en) * | 2019-01-31 | 2019-05-14 | 南京邮电大学 | A kind of more form method for traffic sign detection based on deep learning |
Non-Patent Citations (3)
| Title |
|---|
| ANDREY SOLOVIEV等: "Tight Coupling of GPS, Laser Scanner, and Inertial Measurements for Navigation in Urban Environments", 《POSITION，LOCATION AND NAVIGATION SYMPOSIUM》 * |
| 唐智威: "基于视觉的无人驾驶汽车研究综述", 《制造业自动化》 * |
| 戴燕玲: "基于激光雷达的无人驾驶汽车道路、交通标志与障碍物识别方法", 《无线互联科技》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113465608A (en) * | 2021-07-22 | 2021-10-01 | 清华大学苏州汽车研究院（吴江） | Calibration method and system for roadside sensor |
| CN113465608B (en) * | 2021-07-22 | 2024-05-03 | 清华大学苏州汽车研究院（吴江） | Road side sensor calibration method and system |
| CN115164912A (en) * | 2022-06-24 | 2022-10-11 | 宁波均胜智能汽车技术研究院有限公司 | Vehicle position positioning method and device and readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110135387B (en) | 2021-03-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Amini et al. | Variational end-to-end navigation and localization | |
| US11462023B2 (en) | Systems and methods for 3D object detection | |
| US11333506B2 (en) | Computer-vision based positioning for augmented reality navigation | |
| Li et al. | Springrobot: A prototype autonomous vehicle and its algorithms for lane detection | |
| Kim et al. | Robust lane detection based on convolutional neural network and random sample consensus | |
| CN109166140B (en) | Vehicle motion track estimation method and system based on multi-line laser radar | |
| Huang et al. | Robust inter-vehicle distance estimation method based on monocular vision | |
| CN109945858A (en) | It parks the multi-sensor fusion localization method of Driving Scene for low speed | |
| CN111178253A (en) | Visual perception method and device for automatic driving, computer equipment and storage medium | |
| JP6785985B2 (en) | Imaging system | |
| CN110119698A (en) | For determining the method, apparatus, equipment and storage medium of Obj State | |
| CN111256693B (en) | Pose change calculation method and vehicle-mounted terminal | |
| CN110148177A (en) | For determining the method, apparatus of the attitude angle of camera, calculating equipment, computer readable storage medium and acquisition entity | |
| Stone et al. | Skyline-based localisation for aggressively manoeuvring robots using UV sensors and spherical harmonics | |
| KR20210137893A (en) | Method and system for determining position of vehicle | |
| US20210233307A1 (en) | Landmark location reconstruction in autonomous machine applications | |
| CN110487286A (en) | It is a kind of to project the robot pose determining method merged with laser point cloud based on point feature | |
| CN111089598B (en) | Vehicle-mounted lane-level real-time map matching method based on ICCIU | |
| Zhao et al. | Environmental perception and sensor data fusion for unmanned ground vehicle | |
| CN110135387A (en) | A kind of image rapid identification method based on sensor fusion | |
| CN114048536A (en) | Road structure prediction and target detection method based on multitask neural network | |
| CN113743163A (en) | Traffic target recognition model training method, traffic target positioning method and device | |
| CN116503602A (en) | Unstructured environment three-dimensional point cloud semantic segmentation method based on multi-level edge enhancement | |
| CN107945212A (en) | Infrared small and weak Detection of Moving Objects based on inertial navigation information auxiliary and background subtraction | |
| CN114821517A (en) | Method and system for learning neural networks to determine vehicle poses in an environment |
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 | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230808 Address after: Room 410, 4th Floor, No. 4, Yard 19, Huangping Road, Changping District, Beijing 102200 Patentee after: Beijing Kunpeng Borui Technology Co.,Ltd. Address before: 100039 times Instruments Co., Ltd., Ministry of space, No. 50, Yongding Road, Haidian District, Beijing Patentee before: Li Ziyue |