CN103488973B - Vehicle brand recognition methods and system based on image - Google Patents
Vehicle brand recognition methods and system based on image Download PDFInfo
- Publication number
- CN103488973B CN103488973B CN201310416016.6A CN201310416016A CN103488973B CN 103488973 B CN103488973 B CN 103488973B CN 201310416016 A CN201310416016 A CN 201310416016A CN 103488973 B CN103488973 B CN 103488973B
- Authority
- CN
- China
- Prior art keywords
- vehicle
- image
- brand
- car plate
- characteristic vector
- 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.)
- Active
Links
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a kind of vehicle brand recognition methods based on image and system.This method and system can be based on carrying out multi-dimensional position detection in the picture quickly recognizing vehicle brand, and verification and measurement ratio is high while reporting by mistake less.
Description
Technical field
Recognized the present invention relates to vehicle identification, more particularly to vehicle brand.
Background technology
In the intelligent transportation system in city, vehicle brand identification is an important part.It can know automatically
The brand message of other vehicle, such as certain car is popular, Ford or liberation.Vehicle brand identification traffic control system,
Public security system and business system are all widely used., can be according to different type vehicle in road network in traffic control system
Distribution and closeness, carry out different signal lamp regulations, the configuration of traffic police's police strength etc.；In public security system, most directly application
It is the detection of fake license plate vehicle.If the brand that certain car is identified and its car plate when recording inconsistent, are somebody's turn to do in public security system
Vehicle is possible to be fake-licensed car；In business system, when the vehicle brand distribution situation for understanding some region, advertiser's is wide
Accusing dispensing can be more targeted, and effect becomes apparent from.
Vehicle brand identification is in the prior art without ripe effective method.Similar work is mainly vehicle-logo recognition and car
Type is recognized：
Vehicle-logo recognition obtains vehicle brand by the logo on car body in traffic video or traffic image classify to be believed
Breath.This method has two：1) it requires that the definition of video or image is high, and logo is high-visible, but friendship at this stage
Intervisibility frequency/picture quality can not meet requirement；2) premise of logo classification is logo detection, that is, detects the position of logo in the picture
Put.Current logo detection technique is also immature, and verification and measurement ratio is low while wrong report is more, so vehicle-logo recognition can not also be obtained in reality
To application.Herein, wrong report refers to the non-logo position for being mistakenly considered logo.
Vehicle cab recognition has distinguished the species model of vehicle, such as car, buggy, car and lorry etc..Functionally
See, vehicle cab recognition is the subproblem of brand recognition.In the main method of vehicle cab recognition technology, induction coil measurement obtains most wide
General application.Induction coil measurement is to utilize electromagnetic induction principle：When a car by when, the inductance of coil changes, together
When produce certain waveform.Vehicle information is obtained by the classification to waveform.Although method based on induction coil has cost
Cheap benefit, but its problem is also apparent from：1）Inconvenience is installed, it is necessary to destroy road surface；2）Wave noise is big, and discrimination is low；
3）Maintenance cost is high.
Accordingly, it would be desirable to which one kind can quickly recognize vehicle brand, and inspection based on multi-dimensional position detection is carried out in the picture
Survey rate is high to report less method and system by mistake simultaneously.
The content of the invention
The invention provides a kind of vehicle brand recognition methods based on image, this method includes：Detect car plate in input
Position in image；The detected car plate position of correction；Multiple vehicle part positions are calculated according to calibrated car plate position；
Each extraction feature vector from the plurality of vehicle part position；And the characteristic vector extracted is classified and exported
The brand message of vehicle.
In one embodiment of the vehicle brand recognition methods based on image of the present invention, vehicle part position is to utilize
The relation of vehicle part position and car plate position is calculated.
In one embodiment of the vehicle brand recognition methods based on image of the present invention, the plurality of vehicle part position
It is the position selected in the headlight from the vehicle, fog lamp, air grid, rearview mirror, rain brush, logo, bumper etc..
In one embodiment of the vehicle brand recognition methods based on image of the present invention, this feature vector includes profile
Feature and shape facility.
The present invention the vehicle brand recognition methods based on image one embodiment in, the feature that this pair is extracted to
The step of amount is classified includes：Judging the shape facility, whether the standard shape feature with being stored matches；If it does,
Confidence level of the vehicle relative to each brand is then calculated based on the resemblance；The confidence level highest brand is defined as this
The brand of vehicle.
In one embodiment of the vehicle brand recognition methods based on image of the present invention, the input picture can be car
The image of headstock, or the vehicle tailstock image.
The invention provides a kind of vehicle brand identifying system based on image, the system includes：Car plate detection module, is used
In the position of detection car plate in the input image；VLP correction module, for correcting detected car plate position；Vehicle part
Position acquisition module, for calculating multiple vehicle part positions according to calibrated car plate position；Feature extraction module, for from
Each extraction feature vector of the plurality of vehicle part position；And brand sort module, for the feature to being extracted to
The brand message of vehicle is classified and exported to amount.
In one embodiment of the vehicle brand identifying system based on image of the present invention, the plurality of vehicle part position
It is the position selected in the headlight from the vehicle, fog lamp, air grid, rearview mirror, rain brush, logo, bumper etc..
In one embodiment of the vehicle brand identifying system based on image of the present invention, this feature vector includes profile
Feature and shape facility.
In one embodiment of the vehicle brand identifying system based on image of the present invention, the brand sort module enters one
Walking is used for：Judging the shape facility, whether the standard shape feature with being stored matches；If it does, then special based on the profile
Levy and calculate confidence level of the vehicle relative to each brand；The confidence level highest brand is defined as to the brand of the vehicle.
In one embodiment of the vehicle brand identifying system based on image of the present invention, the input picture can be car
The image of headstock, or the vehicle tailstock image.
Thus, the invention provides a set of ripe effective car plate brand recognition scheme.It can both handle general cleer and peaceful height
Clear video and image, can handle the traffic video and image under complex scene again.The program is to vehicle in video and image
Travel direction do not require, you can processing front part of vehicle, can also handle vehicle tail.Its cost of implementation is low, reliability is high, leads to
Common computer, video camera is crossed to achieve that, and recognition speed is fast, can just be reached using common computer in video analysis
To Real time identification.
Brief description of the drawings
Fig. 1 shows the flowchart overview of the vehicle brand recognition methods according to embodiments of the present invention based on image；
Fig. 2 shows car plate detection schematic diagram.
Fig. 3 shows car plate alignment model schematic diagram.
Fig. 4 shows that car plate matches schematic diagram.
Fig. 5 shows that schematic diagram is extracted in vehicle part position.
Fig. 6 shows brand classification schematic diagram.
Fig. 7 shows the matching classification results figure for vehicle headstock.
Fig. 8 shows the matching classification results figure for the vehicle tailstock.
Fig. 9 shows the block diagram of the vehicle brand identifying system according to embodiments of the present invention based on image.
Embodiment
In the present invention, vehicle brand identification is carried out based on image.Its input is certain frame of single image or video.
Fig. 1 shows the flowchart overview of the vehicle brand recognition methods according to embodiments of the present invention based on image.
The vehicle brand recognition methods based on image of the present invention comprises the following steps：
(a) car plate detection：In step 101, the position of detection car plate in the input image.In this step, input is schemed
As doing car plate detection, its output is the position of car plate in the picture.In one embodiment, position includes car plate in image
Coordinate, car plate height and car plate width.
Piece image can export multiple car plate positions.As shown in Fig. 2 car plate detection schematic diagram, the width image is outputed
3 car plate positions, represent car plate position by rectangle frame respectively.The precision for the car plate position that vehicle detection is obtained not enough it is accurate simultaneously
Also it is unstable, it may appear that the phenomenon such as bigger than normal/less than normal/left and right translation.
There are various schemes in current car plate detection, respectively there is its advantage and disadvantage.In the present embodiment, the step is employed
AdaBoost+Haar Feature scheme.It the advantage is that：(can handle various complex scenes) applied widely, performance is high,
Speed meets demand simultaneously.
(b) car plate position correction：In step 102, the detected car plate position of correction.In this step, car plate is examined
The car plate position measured is corrected.Its output is the exact position of car plate in the picture.Car plate position correction step can be in car
It is corrected on the basis of board detection, obtains car plate exact position.Exact position is the guarantee of car plate brand recognition performance.
In the present embodiment, the step employs the algorithm of binaryzation, and characters on license plate is come out from car plate background segment, from
And obtain the exact position of car plate.The Binarization methods advantage is：Speed is fast, applied widely.
In the prior art, two-dimentional Car license recognition is known with localization method.And in the present invention, it is necessary to based on figure
Picture, is quickly recognized to vehicle brand.Now, the identification and positioning for only carrying out two dimension are inadequate；It is desirable that carrying out
The identification and positioning of multidimensional.
Therefore, in one embodiment of the invention, the mode for employing Model Matching carries out the car plate position school of vehicle
Just.By taking headstock as an example, the model of headstock by N number of vehicle part module composition, each vehicle part module be expressed as (xi, yi,
Fi), wherein (xi, yi) represents the position of vehicle part module, fi is the characteristic vector of the vehicle part module.SIFT description
Because it is quick and descriptive power characteristic such as by force, characteristic vector is used as.The example of model is shown in Fig. 3.
In figure 3, each circle in left side represents a vehicle part module on car body, and right side is single unit vehicle part mould
The SIFT schematic diagrames of block.
Here, using the method for multilevel matching, Optimum Matching is thus found on image.First, to the every of image
Individual position (x, y) all extracts SIFT feature vector.Because the car light of different brands/vehicle or the area of logo are inconsistent, therefore
Each position needs to extract the characteristic vector of different scale.Then, optimal matching is found in image.After solution terminates, it can obtain
To the position of each vehicle part module and overall similarity.Tailstock model is matched using identical method.Finally, by sentencing
The overall similarity of disconnected headstock/tailstock, can determine whether vehicle image is headstock or the tailstock.
In a word, by car plate position correction, it can obtain the direction of vehicle（Headstock/the tailstock）With each vehicle part on car body
The result of module, concrete outcome is shown in Fig. 4.
(c) vehicle part position acquisition：In step 103, multiple vehicle part positions are calculated according to calibrated car plate position
Put.In this step, vehicle part position is calculated according to the car plate position of car plate alignment output, it is vehicle part position that it, which is exported,
Put.This is accomplished by being calculated using the relation of vehicle part position and car plate position.
The position of each vehicle part module in car plate position correction step, what image was normalized is handled
To standard size.The position of each vehicle part module such as headlight, fog lamp, air grid is obtained on standard-sized image, is used
One rectangle frame is represented (x, y, w, h).Concrete outcome is shown in Fig. 5.
(d) feature extraction：In step 104, each extraction feature vector from multiple vehicle part positions.In the step
In, the extraction feature vector in vehicle part position, its output is characteristic vector.Characteristic vector describes the vehicle part band of position
Various characteristics, but can be handled simultaneously by computerized algorithm.
On standard-sized image, each vehicle part module extraction feature vector.Characteristic vector is by two parts
Constitute, a part is resemblance；Another part is shape facility.The exemplary plot of characteristic vector is shown in Fig. 6.
Shape facility in the present embodiment uses HoG (Histogram of Gradient) characteristic vector.Main cause
It is：On shape, texture and the position of the vehicle main distinction part of different brands, such as the shape of car light and position.
HoG can preferably describe this class feature, while having the fireballing advantage of extraction.
And the resemblance in the present embodiment uses LBP (Local Binary Pattern) characteristic vector.
Then by the characteristic vector of each vehicle part module be linked to be a total characteristic vector (f1, f2, f3 ...) (x,
y,w,h)。
(e) brand is classified：In step 105, classified and exported the brand message of vehicle to the characteristic vector extracted.
In this step, brand classification includes following sub-step：1) judge the shape facility whether with the mark that is stored
Quasi- shape facility matches；2) if it does, then calculating vehicle the putting relative to each brand based on the resemblance
Reliability；3) the confidence level highest brand is defined as to the brand of the vehicle.
In step 1）In, judge whether shape facility meets the requirements.For the vehicle of different brands, each car is previously stored
The positions and dimensions of the standard shape feature, i.e. the vehicle part module of component models.For each brand, shape is calculated special
The difference sought peace between " standard shape feature ".If the difference is less than the threshold value being previously set, judgement is likely to be the product
The vehicle of board, the i.e. shape facility match with the standard shape feature stored, into step 2), otherwise interrupt.
In step 2）In, each brand is used as the grader based on resemblance by the use of SVM classifier in advance.Utilize classification
Device can obtain the confidence level of the brand.
In step 3）In, for all brands for obtaining confidence level, selection confidence level highest brand is used as final knot
Really.Final result is shown in Fig. 7 and Fig. 8.
Fig. 7 shows the matching classification results for vehicle headstock.And Fig. 8 shows the matching classification knot for the vehicle tailstock
Really.From figure 7, it is seen that for many vehicles in same image on different tracks, can be based on the matching point for vehicle headstock
Class, quickly and reliably obtains vehicle brand.As can be seen from Fig. 8, can for many vehicles on same track in same image
Based on the matching classification for the vehicle tailstock, vehicle brand is quickly and reliably obtained.
It is the typical problem in machine learning to carry out classification to characteristic vector, has a large amount of ready-made algorithms to solve.This reality
Apply the method that example uses multiclass linear SVM.Main cause is：Speed is fast, performance is high.
Fig. 9 shows the block diagram of the vehicle brand identifying system according to embodiments of the present invention based on image.
The vehicle brand identifying system based on image of the present invention includes：Car plate detection module, for detecting car plate defeated
Enter the position in image；VLP correction module, for correcting detected car plate position；Vehicle part position acquisition module,
For calculating multiple vehicle part positions according to calibrated car plate position；Feature extraction module, for from the multiple vehicle
Each extraction feature vector of component locations；And brand sort module, for classifying to the characteristic vector extracted
And export the brand message of vehicle.
Thus, vehicle product quickly can be recognized based on multi-dimensional position detection is carried out in the picture the invention provides one kind
Board, and verification and measurement ratio is high while reporting less method and system by mistake.This method and system can both handle the video of general cleer and peaceful high definition
And image, the traffic video and image under complex scene can be handled again.Traveling side of the program to vehicle in video and image
To not requiring, you can processing front part of vehicle, vehicle tail can be also handled.Its cost of implementation is low, reliability is high, by commonly counting
Calculation machine, video camera are achieved that, and recognition speed is fast, and real-time knowledge can be just reached using common computer in video analysis
Not.
Vehicle part position is the position of all parts of vehicle, these parts such as headlight, fog lamp, air grid, backsight
Mirror, rain brush, logo, bumper etc..
Example system described herein only applies to the example of some realizations, and is not intended to can be achieved herein
Described process, the environment of component and feature, the use scope or function of architecture and framework propose any limitation.Cause
This, realization herein can be used for numerous environment or architecture, and can be in universal or special computing system or with processing
Realized in the other equipment of ability.In general, software, hardware all can be used in any function being described with reference to the drawings（For example, solid
Determine logic circuit）Or these combinations realized are realized.Term used herein " module " or " component " are typicallyed represent can quilt
It is configured to realize the combination of the software, hardware or software and hardware of predetermined function.For example, in the case of a software implementation, term
" module " or " component " can be represented when in one or more processing equipments（For example, CPU or processor）Perform and specify during upper execution
Task or the program code of operation.Program code can be stored in one or more computer readable memory devices or other meters
In calculation machine storage device.Thus, process described herein, component and module can be realized by computer program product.
Although describing the present invention according to several embodiments, those skilled in the art will recognize the invention is not restricted to
Described embodiment, but the present invention can be implemented with the modifications and changes in spirit and scope of the appended claims, therefore
Specification is considered as illustrative and not restrictive.
Claims (10)
1. a kind of vehicle brand recognition methods based on image, methods described includes：
Detect the position of car plate in the input image；
Based on multidimensional information, detected car plate position is corrected by the way of Model Matching, the essence of car plate is obtained
True position；
Multiple vehicle part positions are calculated according to calibrated car plate exact position, each of which vehicle part be expressed as (xi,
Yi, fi), wherein (xi, yi) represents the position of the vehicle part, fi is the characteristic vector of the vehicle part；
Each extraction feature vector from the multiple vehicle part position, the characteristic vector includes resemblance and shape
Feature, the shape facility uses HoG characteristic vector, and the resemblance uses LBP characteristic vector；
The characteristic vector of each vehicle part is linked to be to total characteristic vector (f1, f2, f3 ...) (x, y, w, h), w is width, h
For height；And
Total characteristic vector is classified and the brand message of vehicle is exported.
2. the vehicle brand recognition methods as claimed in claim 1 based on image, it is characterised in that the vehicle part position
It is to be calculated using the relation of the vehicle part position and the car plate exact position.
3. the vehicle brand recognition methods as claimed in claim 1 based on image, it is characterised in that the multiple vehicle part
The position that position is selected in being the headlight from the vehicle, fog lamp, air grid, rearview mirror, rain brush, logo, bumper.
4. the vehicle brand recognition methods as claimed in claim 1 based on image, it is characterised in that described to total spy
Levying the step of vector is classified includes：
Judging the shape facility, whether the standard shape feature with being stored matches；
If it does, then calculating confidence level of the vehicle relative to each brand based on the resemblance；
The confidence level highest brand is defined as to the brand of the vehicle.
5. the vehicle brand recognition methods as claimed in claim 1 based on image, it is characterised in that can handle in video image
The vehicle of different travel directions, the input picture can be the image of vehicle headstock, or the vehicle tailstock image.
6. a kind of vehicle brand identifying system based on image, the system includes：
Car plate detection module, for detecting the position of car plate in the input image；
VLP correction module, for based on multidimensional information, being carried out by the way of Model Matching to detected car plate position
Correction, obtains the exact position of car plate；
Vehicle part position acquisition module, for calculating multiple vehicle part positions according to calibrated car plate exact position, its
In each vehicle part be expressed as (xi, yi, fi), wherein (xi, yi) represents the position of the vehicle part, fi is the vehicle
The characteristic vector of part；
Feature extraction module, for each extraction feature vector from the multiple vehicle part position, the characteristic vector
Including resemblance and shape facility, the shape facility uses HoG characteristic vector, and the resemblance uses LBP spy
Vector is levied, and the characteristic vector of each vehicle part is linked to be to total characteristic vector (f1, f2, f3 ...) (x, y, w, h), w is
Width, h is height；And
Brand sort module, for being classified and being exported the brand message of vehicle to total characteristic vector.
7. the vehicle brand identifying system as claimed in claim 6 based on image, it is characterised in that the vehicle part position
It is to be calculated using the relation of the vehicle part position and the car plate exact position.
8. the vehicle brand identifying system as claimed in claim 6 based on image, it is characterised in that the multiple vehicle part
The position that position is selected in being the headlight from the vehicle, fog lamp, air grid, rearview mirror, rain brush, logo, bumper.
9. the vehicle brand identifying system as claimed in claim 6 based on image, it is characterised in that the brand sort module
It is further used for：
Judging the shape facility, whether the standard shape feature with being stored matches；
If it does, then calculating confidence level of the vehicle relative to each brand based on the resemblance；
The confidence level highest brand is defined as to the brand of the vehicle.
10. the vehicle brand identifying system as claimed in claim 6 based on image, it is characterised in that video image can be handled
The vehicle of middle different travel directions, the input picture can be the image of vehicle headstock, or the vehicle tailstock image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310416016.6A CN103488973B (en) | 2013-09-12 | 2013-09-12 | Vehicle brand recognition methods and system based on image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310416016.6A CN103488973B (en) | 2013-09-12 | 2013-09-12 | Vehicle brand recognition methods and system based on image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103488973A CN103488973A (en) | 2014-01-01 |
| CN103488973B true CN103488973B (en) | 2017-10-20 |
Family
ID=49829182
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310416016.6A Active CN103488973B (en) | 2013-09-12 | 2013-09-12 | Vehicle brand recognition methods and system based on image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103488973B (en) |
Families Citing this family (27)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105279476A (en) * | 2014-07-15 | 2016-01-27 | 贺江涛 | Vehicle face recognition method and device based on multiple features |
| CN105279475A (en) * | 2014-07-15 | 2016-01-27 | 贺江涛 | Fake-licensed vehicle identification method and apparatus based on vehicle identity recognition |
| CN105426899B (en) * | 2014-09-19 | 2019-11-08 | 腾讯科技（北京）有限公司 | Vehicle identification method, device and client |
| CN104239531B (en) * | 2014-09-19 | 2017-09-26 | 上海依图网络科技有限公司 | A kind of precise alignment method based on local visual feature |
| CN104392611B (en) * | 2014-11-17 | 2016-08-24 | 英华达（南京）科技有限公司 | A kind of identification high price automobile method and system |
| CN104408431B (en) * | 2014-11-28 | 2017-12-26 | 江苏物联网研究发展中心 | Car money recognition methods under traffic monitoring |
| CN104778444B (en) * | 2015-03-10 | 2018-01-16 | 公安部交通管理科学研究所 | The appearance features analysis method of vehicle image under road scene |
| CN105160340A (en) * | 2015-08-31 | 2015-12-16 | 桂林电子科技大学 | Vehicle brand identification system and method |
| CN105787437B (en) * | 2016-02-03 | 2017-04-05 | 东南大学 | A kind of vehicle brand kind identification method based on cascade integrated classifier |
| CN107092855A (en) * | 2016-02-18 | 2017-08-25 | 日本电气株式会社 | Vehicle part recognition methods and equipment, vehicle identification method and equipment |
| CN105551261A (en) * | 2016-03-04 | 2016-05-04 | 博康智能网络科技股份有限公司 | False-license-plate vehicle detection method and system |
| CN105740855A (en) * | 2016-03-24 | 2016-07-06 | 博康智能信息技术有限公司 | Front and rear license plate detection and recognition method based on deep learning |
| CN107506759A (en) * | 2016-06-14 | 2017-12-22 | 杭州海康威视数字技术股份有限公司 | A kind of motor vehicle brand identification method and device |
| CN107545216A (en) * | 2016-06-29 | 2018-01-05 | 深圳市格视智能科技有限公司 | A kind of vehicle identification method available for power-line patrolling |
| CN107016390B (en) * | 2017-04-11 | 2019-11-12 | 华中科技大学 | A kind of vehicle part detection method and system based on relative position |
| CN107133574B (en) * | 2017-04-12 | 2022-06-17 | 浙江大华技术股份有限公司 | Vehicle feature identification method and device |
| CN108171248A (en) * | 2017-12-29 | 2018-06-15 | 武汉璞华大数据技术有限公司 | A kind of method, apparatus and equipment for identifying train model |
| CN108519675B (en) * | 2018-03-25 | 2021-05-07 | 东莞市华睿电子科技有限公司 | Scene display method combining head-mounted display equipment and unmanned vehicle |
| CN110895692B (en) * | 2018-09-13 | 2023-04-07 | 浙江宇视科技有限公司 | Vehicle brand identification method and device and readable storage medium |
| CN112036421B (en) * | 2019-05-16 | 2024-07-02 | 北京搜狗科技发展有限公司 | Image processing method and device and electronic equipment |
| CN110491133B (en) * | 2019-08-08 | 2020-10-16 | 善泊科技(珠海)有限公司 | Vehicle information correction system and method based on confidence |
| CN110991255B (en) * | 2019-11-11 | 2023-09-08 | 智慧互通科技股份有限公司 | Method for detecting fake-licensed car based on deep learning algorithm |
| CN111242951B (en) * | 2020-01-08 | 2024-10-01 | 上海眼控科技股份有限公司 | Vehicle detection method, device, computer equipment and storage medium |
| US11386680B2 (en) | 2020-03-28 | 2022-07-12 | Wipro Limited | System and method of identifying vehicle brand and model |
| CN111507342B (en) * | 2020-04-21 | 2023-10-10 | 浙江大华技术股份有限公司 | Image processing method, device, system and storage medium |
| CN111738228A (en) * | 2020-08-04 | 2020-10-02 | 杭州智诚惠通科技有限公司 | Multi-view vehicle feature matching method for hypermetrological evidence chain verification |
| CN112102408B (en) * | 2020-09-09 | 2024-07-23 | 东软睿驰汽车技术（沈阳）有限公司 | Method and device for correcting monocular vision scale and automatic driving automobile |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101630361A (en) * | 2008-12-30 | 2010-01-20 | 北京邮电大学 | Plate number, body color and mark identification-based equipment and plate number, body color and mark identification-based method for identifying fake plate vehicles |
| CN102708691A (en) * | 2011-12-26 | 2012-10-03 | 南京信息工程大学 | False license plate identification method based on matching between license plate and automobile type |
| CN103065143A (en) * | 2012-12-30 | 2013-04-24 | 信帧电子技术（北京）有限公司 | Automobile logo identification method and device |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8917904B2 (en) * | 2008-04-24 | 2014-12-23 | GM Global Technology Operations LLC | Vehicle clear path detection |
-
2013
- 2013-09-12 CN CN201310416016.6A patent/CN103488973B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101630361A (en) * | 2008-12-30 | 2010-01-20 | 北京邮电大学 | Plate number, body color and mark identification-based equipment and plate number, body color and mark identification-based method for identifying fake plate vehicles |
| CN102708691A (en) * | 2011-12-26 | 2012-10-03 | 南京信息工程大学 | False license plate identification method based on matching between license plate and automobile type |
| CN103065143A (en) * | 2012-12-30 | 2013-04-24 | 信帧电子技术（北京）有限公司 | Automobile logo identification method and device |
Non-Patent Citations (2)
| Title |
|---|
| 《基于混合特征的车牌识别技术研究》;赵兵;《中国优秀硕士学位论文全文数据库》;20090715;第2.3节 * |
| 基于图像信息与模糊神经网络的特征识别技术及其应用;薛天以;《中国优秀硕士学位论文全文数据库》;20110915(第9期);第2.1.2节最后一段，第2.1.3节，第2.2.1节，图2.12 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103488973A (en) | 2014-01-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103488973B (en) | Vehicle brand recognition methods and system based on image | |
| Satzoda et al. | Multipart vehicle detection using symmetry-derived analysis and active learning | |
| Cheon et al. | Vision-based vehicle detection system with consideration of the detecting location | |
| He et al. | Recognition of car makes and models from a single traffic-camera image | |
| US10025998B1 (en) | Object detection using candidate object alignment | |
| KR101596299B1 (en) | Apparatus and Method for recognizing traffic sign board | |
| Kuang et al. | Feature selection based on tensor decomposition and object proposal for night-time multiclass vehicle detection | |
| CN105894819B (en) | A kind of fake-licensed car recognition methods based on secondary verification | |
| CN110135377B (en) | Method and device for detecting motion state of object in vehicle-road cooperation and server | |
| US20230245466A1 (en) | Vehicle Lidar System and Object Classification Method Therewith | |
| CN105354533A (en) | Bag-of-word model based vehicle type identification method for unlicensed vehicle at gate | |
| Deng et al. | Detection and recognition of traffic planar objects using colorized laser scan and perspective distortion rectification | |
| Gao et al. | On‐line vehicle detection at nighttime‐based tail‐light pairing with saliency detection in the multi‐lane intersection | |
| CN110766009A (en) | Tail plate identification method and device and computer readable storage medium | |
| Chang et al. | Stereo-based object detection, classi? cation, and quantitative evaluation with automotive applications | |
| Ali et al. | On-road vehicle detection using support vector machine and decision tree classifications | |
| CN111161542B (en) | Vehicle identification method and device | |
| Nassu et al. | Automatic recognition of railway signs using sift features | |
| Hsu et al. | Developing an on-road obstacle detection system using monovision | |
| Lin et al. | Incorporating appearance and edge features for vehicle detection in the blind-spot area | |
| Bota et al. | A framework for object detection, tracking and classification in urban traffic scenarios using stereovision | |
| Prahara et al. | Vehicle pose estimation for vehicle detection and tracking based on road direction | |
| Rao et al. | Object Sub-Categorization and Common Framework Method using Iterative AdaBoost for Rapid Detection of Multiple Objects | |
| Jin et al. | Method for pedestrian detection using ground plane constraint based on vision sensor | |
| Shin et al. | Vehicle classification by road lane detection and model fitting using a surveillance camera |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |