CN104715614B - Method for capturing suspected fake-licensed cars - Google Patents
Method for capturing suspected fake-licensed cars Download PDFInfo
- Publication number
- CN104715614B CN104715614B CN201510102368.3A CN201510102368A CN104715614B CN 104715614 B CN104715614 B CN 104715614B CN 201510102368 A CN201510102368 A CN 201510102368A CN 104715614 B CN104715614 B CN 104715614B
- Authority
- CN
- China
- Prior art keywords
- fake
- vehicle
- suspicion
- car
- storehouse
- 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
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/017—Detecting movement of traffic to be counted or controlled identifying vehicles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/146—Aligning or centring of the image pick-up or image-field
- G06V30/1475—Inclination or skew detection or correction of characters or of image to be recognised
- G06V30/1478—Inclination or skew detection or correction of characters or of image to be recognised of characters or characters lines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/148—Segmentation of character regions
- G06V30/153—Segmentation of character regions using recognition of characters or words
Abstract
The invention relates to a method for capturing suspected fake-licensed cars. The method comprises the following steps that 1, the license plate number and the brand information of a car are identified, a primary judgment result is correspondingly obtained according to the license plate number and the brand information, and if the license plate number does not correspond to the brand information, the car is added to a fake-licensed car suspicion library; 2, the cars added to the fake-licensed car suspicion library are accurately judged, and misreported cars are eliminated; 3, the number of times of appearing of each car in the fake-licensed car suspicion library is calculated, when the number of times of appearing of one car exceeds a preset value K, an alarm is given out. Compared with the prior art, the method has the advantages that the speed is high, and the misinformation rate is low.
Description
Technical field
The present invention relates to technical field of intelligent traffic, especially relate to a kind of suspicion fake-licensed car catching method.
Background technology
Fake-licensed car fingerstall escapes the vehicle of punishment with the number-plate number of other vehicles.Can be escaped by fake-licensed car and be made
People's lives and properties and public safety are caused great threat by the serious consequence becoming and the expense turned over, so always
The object of traffic monitoring department key point strike, and forbidded strictly by country.Therefore, flow through in car data in magnanimity traffic and actively send out
Now with identification fake-licensed car, not only investigate significant to traffic police active forewarning with afterwards, even more to real car owner with
The protection of fake-licensed car accident victim's interests.
The catching method of existing fake-licensed car vehicle, the relation according to the time difference occurring before and after vehicle and distance, sentences mostly
Whether the appearance of disconnected vehicle meets logic, and the method requires doubtful fake-licensed car and true car, and (or short period in front and back) occurs simultaneously,
Qualificationss are strict, carry out inspection dynamics with the method inadequate.
The catching method of the fake-licensed car vehicle based on smart city of Application No. 201310034242.8, it is by electricity
The data crossing car data storehouse of subcard button carries out gridding classification, and the data that doubtful fake-licensed car is occurred is screened, and then catches
The method obtaining fake license plate vehicle.The technical problem that this method exists is: for vehicle data obtains needs according to certain when
Between sequence obtaining, and then data pair gridded data will be realized generate containing realizing sequence grid track, this will sentence for actual
In fixed, it is difficult to accomplish, because some fake license plate vehicle probably enter in this net region in some time point being stopped,
Perhaps in the retention period, changed license plate number, therefore, this time series cannot be continuous, and then cannot realize to this same car plate
, therefore there is the possibility failed to judge in the grid track monitoring of number vehicle.Furthermore, the division for grid is complex, in conjunction with when
Between sequence combination, the back-end data treating capacity of this mode is very huge, and judging efficiency is relatively low.
A kind of fake-licensed car recognition methodss based on hadoop of Application No. 201410100491.7, adopt, will cut first
Subtract the effective car record of crossing after dimension and move in the hbase of hadoop cluster, then obtained using hive same from hbase
The number-plate number occurs in crossing car record and according to the number-plate number and crossing the sequence of car time packet, then just of any two control point
Beginningization is the weighted graph of side right value by control point for vertex set and between any two distance, calculates all control points between any two
Shortest path, its combination of two piecemeal is processed, finally creates multiple threads, supervises after being processed according to piecemeal under fake-licensed car rule
The combination of two of control point concurrently submits to hive task to identify fake-licensed car, and obtains final suspicion deck by correction factor
Car.But the method is complex, calculating speed is slower.
Content of the invention
The purpose of the present invention is exactly to provide to overcome the defect that above-mentioned prior art exists that a kind of speed is fast, rate of false alarm
Low suspicion fake-licensed car catching method.
The purpose of the present invention can be achieved through the following technical solutions:
A kind of suspicion fake-licensed car catching method, comprises the following steps:
1) license plate number of identification vehicle and brand message, it is preliminary that the correspondence according to license plate number and brand message obtains vehicle
Differentiate result, if license plate number and brand message do not correspond to, this vehicle is added fake-licensed car suspicion storehouse；
2) vehicle adding fake-licensed car suspicion storehouse is accurately differentiated, removed the vehicle of wrong report；
3) count the number of times that each vehicle occurs in fake-licensed car suspicion storehouse, when the occurrence number of certain vehicle exceedes setting value k
When send alarm.
Described correspondence according to license plate number and brand message obtain vehicle principium identification result particularly as follows:
Inquired about in government vehicle administration office data base according to license plate number, obtain corresponding accurate brand, will identify that
Brand message is compared with described accurate brand, judges whether license plate number and brand message correspond to.
Described accurate differentiation includes car plate, and accurately differentiation and brand accurately differentiate.
Described car plate accurately differentiate particularly as follows:
A1) rough position of each character according to the license plate number recognition result positioning car trade mark, splits to each character,
Whether each character after judging to split is complete, if so, then execution step a2), if it is not, then corresponding vehicle removal fake-licensed car is disliked
Doubtful storehouse；
A2) calculate the center of each character, judge whether the angle of inclination of each character is more than setting value, if so, then will
Corresponding vehicle removes fake-licensed car suspicion storehouse, if it is not, then execution step a3)；
A3) adopt deep learning algorithm in step a1) identification of the rough position enterprising line character that obtains, obtain each
The recognition confidence of character, whether the recognition confidence of front twoth character of recognition result value ranking is less than given threshold, if so,
Then corresponding vehicle is removed fake-licensed car suspicion storehouse, if it is not, then execution step a4)；
A4) judging whether containing illumination effect, if so, then corresponding vehicle being removed fake-licensed car suspicion storehouse, if it is not, then holding
Row step a5)；
A5) judge whether current car plate is negative and positive board, if so, then corresponding vehicle is removed fake-licensed car suspicion storehouse, if it is not,
Then exit.
Described brand accurately differentiate particularly as follows:
B1) judge whether vehicle on video image is complete, if so, then execution step b2), if it is not, then by corresponding car
Removal fake-licensed car suspicion storehouse；
B2) judge whether brand message recognition result is located in list of common error, if so, then corresponding vehicle is removed
Fake-licensed car suspicion storehouse, if it is not, then execution step b3)；
B3) judge whether the recognition confidence of brand message recognition result is less than setting value, if so, then by corresponding vehicle
Removal fake-licensed car suspicion storehouse, if it is not, then exit.
The vehicle brand that appearance similarity degree is more than 90% is preserved to information in described list of common error.
The value of described k is 3.
Compared with prior art, the invention has the advantages that
(1) inventive algorithm speed is fast, implements simple, tests under battle conditions in many districts and cities, all obtains positive effect.
(2) present invention, on the basis of principium identification, carries out the accurate differentiation of car plate and brand, effectively eliminates to vehicle
Differentiate the situation of mistake, rate of false alarm is low.
(3) present invention accounts for many factors in the accurate differentiation of car plate and brand, effectively increases deck
The accuracy of car identification.
Brief description
Fig. 1 is the schematic flow sheet of the present invention.
Specific embodiment
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings.The present embodiment is with technical solution of the present invention
Premised on implemented, give detailed embodiment and specific operating process, but protection scope of the present invention be not limited to
Following embodiments.
As shown in figure 1, the embodiment of the present invention provides a kind of suspicion fake-licensed car catching method, comprise the following steps:
Step s101: detected by car plate, Car license recognition and brand recognition algorithm identify that the license plate number of vehicle and brand are believed
Breath, the correspondence according to license plate number and brand message obtains vehicle principium identification result, particularly as follows:
Inquired about in government vehicle administration office data base according to license plate number, obtain corresponding accurate brand, will identify that
Brand message is compared with described accurate brand, judges whether license plate number and brand message correspond to, if license plate number and brand letter
Breath does not correspond to, then this vehicle is added fake-licensed car suspicion storehouse.
Step s102: the vehicle adding fake-licensed car suspicion storehouse is accurately differentiated, removes the vehicle of wrong report.
For most of Car license recognition/brand recognition algorithm, the accuracy rate of identification only has 90%, therefore on road surface
The probability that vehicle has 10% can identify mistake, and leading to is considered as fake-licensed car；But the probability of fake-licensed car is less than ten thousand on road surface
/ mono- it is therefore desirable to differentiate to remove wrong report using accurate.Accurately differentiate that including the accurate differentiation of car plate accurately differentiates with brand.
Car license recognition mistake main cause is to block (have soil vestige on car plate, blocked intentionally with characters on license plate by car
Frame shield portions are interval), illumination (too dark or negative and positive board), tilt and similar character etc., therefore, car plate accurately differentiates
Particularly as follows:
A1) the rough position of each character according to the license plate number recognition result positioning car trade mark, can be by the generally recognized calculation
Method obtains this result, and each character is split, and whether each character after judging to split is complete, if so, then execution step a2),
If it is not, then corresponding vehicle is removed fake-licensed car suspicion storehouse；
A2) calculate the center of each character, judge whether the angle of inclination of each character is more than setting value, such as 15 degree, if
It is then corresponding vehicle to be removed fake-licensed car suspicion storehouse, if it is not, then execution step a3)；
A3) adopt deep learning algorithm in step a1) identification of the rough position enterprising line character that obtains, obtain each
The recognition confidence of character, whether the recognition confidence of front twoth character of recognition result value ranking is less than given threshold, if so,
Then corresponding vehicle is removed fake-licensed car suspicion storehouse, if it is not, then execution step a4)；
A4) judging whether containing illumination effect, if so, then corresponding vehicle being removed fake-licensed car suspicion storehouse, if it is not, then holding
Row step a5)；
Illumination effect refers to be judged according to the difference of prospect (character) color and background, if all partially black and difference is little
In certain value, value is between 0.1～0.2 then it is assumed that illumination condition is bad.The difference of prospect (character) color and background calculates
Process is: is averaging color to foreground area and background area, then they is transformed into cielab color space, finally calculates
L2 distance between the two.
A5) judge whether current car plate is negative and positive board, if so, then corresponding vehicle is removed fake-licensed car suspicion storehouse, if it is not,
Then exit.Negative and positive board refers to that background color has notable difference.The determination methods of notable difference are: to background area institute a little
Color carries out dbscan cluster, if the points of the two of maximum classes are above 25%, and their average color are empty in cielab
Between on distance just belong to notable difference more than 0.2.
Brand accurately differentiate particularly as follows:
B1) judge whether vehicle on video image is complete, if so, then execution step b2), if it is not, then by corresponding car
Removal fake-licensed car suspicion storehouse；
B2) judge whether brand message recognition result is located in list of common error, if so, then corresponding vehicle is removed
Fake-licensed car suspicion storehouse, if it is not, then execution step b3)；
Partial domestic vehicle appearance and famous foreign brand are more similar, hence set up list of common error, this frequent fault
The vehicle brand that appearance similarity degree is more than 90% is preserved to information in list.
B3) judge that whether the recognition confidence of brand message recognition result is less than setting value, general value 0 to 1.0 it
Between, if so, then corresponding vehicle is removed fake-licensed car suspicion storehouse, if it is not, then exiting.
Step s103: count the number of times that each vehicle occurs in fake-licensed car suspicion storehouse, when the occurrence number of certain vehicle exceedes
Alarm is sent during setting value k, the corresponding fake-licensed car of capture, the value of k is 3.
Said method speed is fast, and rate of false alarm is low, tests under battle conditions in many districts and cities, all obtains positive effect.
Claims (5)
1. a kind of suspicion fake-licensed car catching method is it is characterised in that comprise the following steps:
1) license plate number of identification vehicle and brand message, the correspondence according to license plate number and brand message obtains vehicle principium identification
As a result, if license plate number and brand message do not correspond to, this vehicle is added fake-licensed car suspicion storehouse；
2) vehicle adding fake-licensed car suspicion storehouse is accurately differentiated, remove the vehicle of wrong report, described accurate differentiation includes car
Board accurately differentiates and brand accurately differentiates；
3) count the number of times that each vehicle occurs in fake-licensed car suspicion storehouse, send out when the occurrence number of certain vehicle exceedes setting value k
Go out alarm；
Described car plate accurately differentiate particularly as follows:
A1) rough position of each character according to the license plate number recognition result positioning car trade mark, splits to each character, judges
Whether each character after segmentation is complete, if so, then execution step a2), if it is not, then corresponding vehicle is removed fake-licensed car suspicion
Storehouse；
A2) calculate the center of each character, judge whether the angle of inclination of each character is more than setting value, if so, then will correspond to
Vehicle removal fake-licensed car suspicion storehouse, if it is not, then execution step a3)；
A3) adopt deep learning algorithm in step a1) identification of the rough position enterprising line character that obtains, obtain each character
Recognition confidence, whether the recognition confidence of front twoth character of recognition result value ranking be less than given threshold, if so, then will
Corresponding vehicle removes fake-licensed car suspicion storehouse, if it is not, then execution step a4)；
A4) judging whether containing illumination effect, if so, then corresponding vehicle being removed fake-licensed car suspicion storehouse, if it is not, then executing step
Rapid a5)；
A5) judging whether current car plate is negative and positive board, if so, then corresponding vehicle being removed fake-licensed car suspicion storehouse, if it is not, then moving back
Go out.
2. suspicion fake-licensed car catching method according to claim 1 is it is characterised in that described believe according to license plate number and brand
Breath correspondence obtain vehicle principium identification result particularly as follows:
Inquired about in government vehicle administration office data base according to license plate number, obtained corresponding accurate brand, the brand that will identify that
Information is compared with described accurate brand, judges whether license plate number and brand message correspond to.
3. suspicion fake-licensed car catching method according to claim 1 is it is characterised in that described brand accurately differentiates specifically
For:
B1) judge whether vehicle on video image is complete, if so, then execution step b2), if it is not, then corresponding vehicle is moved
Go out fake-licensed car suspicion storehouse；
B2) judge whether brand message recognition result is located in list of common error, if so, then corresponding vehicle is removed deck
Car suspicion storehouse, if it is not, then execution step b3)；
B3) judge whether the recognition confidence of brand message recognition result is less than setting value, if so, then corresponding vehicle is removed
Fake-licensed car suspicion storehouse, if it is not, then exit.
4. suspicion fake-licensed car catching method according to claim 3 is it is characterised in that preserve in described list of common error
Appearance similarity degree is had to be more than 90% vehicle brand to information.
5. suspicion fake-licensed car catching method according to claim 1 is it is characterised in that the value of described k is 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510102368.3A CN104715614B (en) | 2015-03-09 | 2015-03-09 | Method for capturing suspected fake-licensed cars |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510102368.3A CN104715614B (en) | 2015-03-09 | 2015-03-09 | Method for capturing suspected fake-licensed cars |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104715614A CN104715614A (en) | 2015-06-17 |
| CN104715614B true CN104715614B (en) | 2017-01-18 |
Family
ID=53414900
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510102368.3A Active CN104715614B (en) | 2015-03-09 | 2015-03-09 | Method for capturing suspected fake-licensed cars |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104715614B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106384513B (en) * | 2016-09-17 | 2019-04-05 | 广东信佰君略科技咨询有限公司 | A kind of fake-licensed car capture system and method based on intelligent transportation |
| CN107067736B (en) * | 2017-04-12 | 2019-10-08 | 安徽超远信息技术有限公司 | Fake-licensed car analysis method and its system based on time road network |
| CN107195181B (en) * | 2017-06-02 | 2019-08-02 | 中通服咨询设计研究院有限公司 | A method of fake-licensed car is identified according to fake-licensed car recognition rule library |
| CN107346435A (en) * | 2017-06-15 | 2017-11-14 | 浙江捷尚视觉科技股份有限公司 | A kind of suspicion fake-licensed car catching method based on vehicle characteristics storehouse |
| CN112116814B (en) * | 2019-06-19 | 2021-06-18 | 杭州海康威视系统技术有限公司 | Abnormal vehicle detection method and device and electronic equipment |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7073063B2 (en) * | 1999-03-27 | 2006-07-04 | Microsoft Corporation | Binding a digital license to a portable device or the like in a digital rights management (DRM) system and checking out/checking in the digital license to/from the portable device or the like |
| US8156339B2 (en) * | 2004-07-21 | 2012-04-10 | Sanyo Electric Co., Ltd. | Method for transmission/reception of contents usage right information in encrypted form, and device thereof |
| CN1928892A (en) * | 2006-09-20 | 2007-03-14 | 王枚 | Method and device for license plate location recognition, vehicle-logo location recognition and vehicle type |
| CN101633346A (en) * | 2008-07-22 | 2010-01-27 | 李金生 | Car licence plate recognition method |
| 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 |
| CN102426786B (en) * | 2011-11-15 | 2014-02-12 | 无锡港湾网络科技有限公司 | Intelligent video analyzing system and method for automatically identifying fake plate vehicle |
| CN102521986B (en) * | 2011-12-05 | 2013-11-06 | 沈阳聚德视频技术有限公司 | Control method for automatic detection system for fake plate vehicle |
| CN103246876B (en) * | 2013-05-10 | 2017-05-10 | 苏州祥益网络科技有限公司 | Image feature comparison based counterfeit vehicle registration plate identification method |
-
2015
- 2015-03-09 CN CN201510102368.3A patent/CN104715614B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN104715614A (en) | 2015-06-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104715614B (en) | Method for capturing suspected fake-licensed cars | |
| Saha et al. | License Plate localization from vehicle images: An edge based multi-stage approach | |
| CN104408475B (en) | A kind of licence plate recognition method and car license recognition equipment | |
| CN106991820B (en) | Illegal vehicle processing method and device | |
| CN103198493B (en) | A kind ofly to merge and the method for tracking target of on-line study based on multiple features self-adaptation | |
| CN105389567A (en) | Group anomaly detection method based on a dense optical flow histogram | |
| CN104992148A (en) | ATM terminal human face key points partially shielding detection method based on random forest | |
| CN105512660A (en) | License number identification method and device | |
| CN107346435A (en) | A kind of suspicion fake-licensed car catching method based on vehicle characteristics storehouse | |
| CN102063614A (en) | Method and device for detecting lost articles in security monitoring | |
| CN105788269A (en) | Unmanned aerial vehicle-based abnormal traffic identification method | |
| CN103839085A (en) | Train carriage abnormal crowd density detection method | |
| CN109344886B (en) | Occlusion number plate distinguishing method based on convolutional neural network | |
| CN102968625A (en) | Ship distinguishing and tracking method based on trail | |
| CN104392239A (en) | License plate identification method and system | |
| CN104244016B (en) | A kind of H264 video contents altering detecting method | |
| CN103902981A (en) | Method and system for identifying license plate characters based on character fusion features | |
| CN109949269A (en) | A kind of detection method and device of railroad train dust cap breakage failure | |
| CN104361366B (en) | A kind of licence plate recognition method and car license recognition equipment | |
| CN108830882B (en) | Video abnormal behavior real-time detection method | |
| CN108830204B (en) | Method for detecting abnormality in target-oriented surveillance video | |
| CN109274945B (en) | Method and system for self-adaptively performing true color restoration on image | |
| CN110826390A (en) | Video data processing method based on face vector characteristics | |
| CN113408364B (en) | Temporary license plate recognition method, system, device and storage medium | |
| CN111062319A (en) | Driver call detection method based on active infrared image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |