CN1694130A - Identification method of mobile number plate based on three-channel parallel artificial nerve network - Google Patents
Identification method of mobile number plate based on three-channel parallel artificial nerve network Download PDFInfo
- Publication number
- CN1694130A CN1694130A CN 200510024593 CN200510024593A CN1694130A CN 1694130 A CN1694130 A CN 1694130A CN 200510024593 CN200510024593 CN 200510024593 CN 200510024593 A CN200510024593 A CN 200510024593A CN 1694130 A CN1694130 A CN 1694130A
- Authority
- CN
- China
- Prior art keywords
- identification
- number plate
- network
- recognition
- net
- 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
Images
Landscapes
- Image Analysis (AREA)
- Character Discrimination (AREA)
Abstract
The invention relates to a floating number plates discriminating method, which is based on the three-road parallel manmade nerve-net. The identify steps are that adopting the video burst mode to catch the video automatically for the collection of video pictures of the moving vehicle, character cutting the part of pictures of the number plates and then being the input signal of the nerve-net; said nerve-net adopts three standard self-adapt vibrate net each of which has its own work, the three nets are the Chinese characters identify net, the English characters identify net and the data identify net, all of which identify the inputted vector signals at one time, and outputs the most similar sort unity respectively, through controlling the property of the Territory value and the identify accurately according to the number plates and filter, at last add a color property of number plates as the output of the discriminating results. Adopting the method of the invention can identify the number plates of vehicle fast and accurately under the largest driving speed of the vehicles.
Description
Technical field:
The present invention relates to a kind of number plate recognition methods, a kind of method of carrying out the identification of target number plate at the vehicle in travelling of more specifically saying so.
Background technology:
Automobile in travelling must be confirmed its identity as the object of being managed and controlling.Current, various intelligent transportation application systems based on the License Plate video identification mainly contain: crossing the make a dash across the red light on-the-spot punishment of automatic recognition system, overspeed of vehicle system, bayonet socket vehicle identification system, automobile automatic fee system, parking lot automated management system, mobile electron policing system etc., License Plate automatic recognition system as said system key sensor device must have fast, accurately, the characteristics obtained easily, so just can reach the practicability purpose.
At present, the License Plate video identification is mainly adopted three kinds of patterns both at home and abroad:
First kind is template matching method, utilizes the comparison of fixed car licence plate template and License Plate character individual element to be identified.This method is simple, but only is fit to the situation of fixed size, horizontal positioned, when car plate motion, rotation or viewing angle change, and mistake identification easily.
Second kind is the proper vector matching method, utilize the stroke feature of each character in the License Plate, input character is decomposed into the set with stroke structure feature, mates with characteristic set in the character repertoire, the characters matched of architectural feature is exactly the recognition result of this input character.This method has been got rid of size, direction changes the interference that brings, but when the character incompleteness, have when stained inapplicable.Based on methods such as the peripheral outline of also having of this technology, projection sequence characteristic matching.
The third method is an artificial neural network fuzzy diagnosis method, utilizes the fuzzy corresponding relation of license plate character pattern and License Plate character object to be identified, adopts artificial neural network technology that the number-plate number is accurately discerned.Because not needing that the pattern of input is done obvious characteristics, do not extract by the common neural network sorter, itself just has feature extraction functions the hidden layer of network, neural network is damaged not too responsive to the incomplete formula feature of pattern information, with traditional mode identification method by comparison, neural network classifier is under ground unrest statistical property condition of unknown, its performance is better, and neural network has higher generalization and adaptive faculty.This way has overcome the drawback of above-mentioned two kinds of methods substantially, has become international and domestic mainstream technology.
Adopt artificial neural network technology mainly to concentrate on aspects such as the simple and easy degree of recognition accuracy, recognition speed, training study and convergence with research to number plate identification, the nerual network technique that can adopt mainly contains BP feedforward network, ART1 self-adaptation concussion network, SA-ART network, BAM-BP network etc., the whole bag of tricks is each has something to recommend him, numeral and English alphabet discrimination can reach 98% substantially, the Chinese character discrimination does not wait from 80%-90%, composite vehicle license plate identification rate is 85%-90%, has reached 95% individually.From concrete applicable cases, at present only relevant application is arranged on the BP network technology, but the BP network exist cycle of training long, the cluster convergence is slow, bad adaptability, may produce extreme-value problem.About the ART network technology, do not find as yet in number plate identification, to carry out practical application, simultaneously, existing ART network has the problem of precision and class diffusion contradiction.
Summary of the invention:
The invention provides a kind of identification method of mobile number plate that allows to discern accurately, rapidly under the movement velocity its license plate number at the vehicle in travelling in maximum based on three-channel parallel artificial nerve network.
The present invention's technical scheme that is adopted of dealing with problems is:
A kind of identification method of mobile number plate based on three-channel parallel artificial nerve network is that identification step is as follows:
1), gather the processing of License Plate video image and signal acquisition:
A) adopt video camera,, adopt the sport video triggering mode to trigger automatically and carry out video capture the collection of moving vehicle video image.
B) the video capture signal is carried out after the Character segmentation input signal as neural network.
2), described neural network adopts the self-adaptation concussion neural network (ART1) of three standards of having one's own work to do, be Chinese Character Recognition network (ART1-1), English alphabet recognition network (ART1-2) and digit recognition network (ART1-3), three standard networks are discerned input signal simultaneously, and its operation steps is:
A, elder generation train respectively three parallel neural networks, make recognition system reach required accuracy of identification and adaptability.
B, input signal is discerned:
1., from described three recognition networks, to exporting three the highest classification individualities of similarity respectively after the input signal identification, character alignment attribute characteristics by predefined License Plate template, individual to described three classifications output, carry out uniqueness according to number plate character field value attribute and select, guarantee that the result of each character bit output is that most probable and similarity are the highest;
2., when similarity does not reach setting value, handle by fault-tolerant mode and to export;
3., when a complete License Plate through filtering, after the double accurate recognition result unanimity, add the number plate color attribute and export as recognition result.
Above-mentioned sport video triggering mode for triggering automatically is, when moving vehicle enters the default virtual detection coil of screen, to cause the variation of video image, recognition system adopts the License Plate graphics template that picture is scanned fast, when finding that possibility number plate zone exists, the automatic mark of system, after the high-velocity scanning of finishing full frame, again analysis and judgement is carried out in a plurality of zones of being found, find out real number plate and cut apart and handle, can finish like this there is the seizure of a plurality of License Plates in single-image and improves capture velocity.
The training of above-mentioned neural network, be to adopt no tutor's training and have the tutor to train to combine, promptly after the three-channel parallel neural network is finished the sampling learning tasks of all categories respectively, Adjustment System accuracy of identification controlling value is closed the newly-increased classification function of ART1 network, proceeds the high precision recognition training, when system prompt has new classification to produce, by artificial selection and classification, continually strengthen accuracy of identification and adaptability, better solved the precision of ART1 network existence and the problem of class diffusion contradiction.
The identification of above-mentioned neural network is to adopt compound method of identification, double identification, and the then output that comes to the same thing, then identification again of difference as a result in the possible identification processing time, through repeatedly identification continuously, is discerned output until identical the finishing of front and back two times result.
Compared with the prior art, beneficial effect of the present invention is embodied in:
The present invention is directed to the characteristics of Chinese automobile licence plate, adopt the identification of three-channel parallel ART1 network class Chinese character, English alphabet and numeral at cognitive phase, carry out quick clustering in the method that the neural metwork training stage adopts no tutor and has the tutor to intervene, adopt thresholding filtering and fault-tolerant technique etc. at the neural network output stage.By the integrated use of many technology, overcome long, slow, the bad adaptability of cluster convergence of BP network training cycle, may produce problems such as extreme value, also overcome the problem of conventional self-adaptation concussion network A RT1 precision and class diffusion contradiction simultaneously.Actual test, the single recognition accuracy is 80% behind the simple exercise, compound discrimination be 90%, 5000 training and fault-tolerant after, dynamic actual recognition accuracy is moved on the road surface can reach 98% (7 characters of one pair of license plate have dislocation erroneous judgement mistake), vehicle movement speed is during less than 160 kilometers/hour, system can accurately discern, but detects target and tester's relative motion, system stability, dependable performance has reached degree of being practical.
Description of drawings:
Fig. 1 is the number plate illustration that the embodiment of the invention is discerned.
Fig. 2 is that the ART1 network signal is handled block diagram.
Fig. 3 is the signal Processing block diagram that the present invention is based on the three-channel parallel neural network.
Embodiment:
A preferred embodiment example of the present invention is: referring to Fig. 1, present embodiment is the mobile License Plate identification way that adopts based on three-channel parallel artificial nerve network, and " Anhui A81770 " discerns to a number plate example; Step is:
1), the collection of number plate video image:
Adopt default virtual coil sport video triggering mode to trigger automatically at video image and carry out video capture, the video capture signal is carried out after the Character segmentation input signal as neural network.
The automatic triggering of sport video triggering mode is specially, when moving vehicle enters the default virtual detection coil of screen camera views, to cause the variation of image, system adopts the License Plate graphics template that picture is scanned fast, when finding that possibility number plate zone exists, the automatic mark of system, after the high-velocity scanning of finishing full frame, again analysis and judgement is carried out according to the pixel corresponding relation of licence plate in a plurality of zones of being found, find out real number plate and cut apart and handle, can finish like this there is the seizure of a plurality of License Plates in single-image and improves capture velocity.
In concrete the enforcement, can adopt high-resolution digital one video camera, finish collection, require resolution of video camera to reach more than 500 tv lines the moving vehicle video image, electronic shutter speed directly uses 1394 digital interfaces to link to each other more than 1/1000 second with computing machine.
After system receives and catches trigger pip, directly screen video memory information is carried out acquisition process, comprise directly single-frame images is carried out the figure image intensifying, image filtering, picture is carried out the model processing, and then image is carried out two-value handle, utilize image histogram broadening technology to strengthen picture contrast, utilize car plate template length and width relation again, automatically carry out the image boundary match search from the bottom up, after finding target, at first carry out coarse localization, continue search simultaneously until finishing, to the suspicious result who searches, proportionate relationship according to licence plate black and white element is tentatively eliminated fake license, last possible licence plate is carried out image to take, the size tilt adjustments, Character segmentation, after the normalized, last with the input signal of vector array form as neural network, carry out the identification of number plate and the processing of data.
2), the identification of number plate:
In neural network, the self-adaptation concussion neural network (ART1) of three standards that employing is had one's own work to do, comprise Chinese Character Recognition network A RT1-1, English alphabet recognition network ART1-2 and digit recognition network A RT1-3, three standard networks are discerned input vector simultaneously, from three classes, export the highest individuality of similarity respectively, the individuality of top three classifications being exported by the character alignment attribute characteristics of predefined License Plate template carries out the uniqueness selection according to number plate character field value attribute, guarantees that the result of each character bit output is that most probable and similarity are the highest; Do not reach setting value in similarity, handle by fault-tolerant mode and export; After the accurate identification of a complete License Plate is finished, add the number plate color attribute and export as recognition result.
Training for system neural network:
In order to solve the problem of issuable newly-increased classification individuality in the intrinsic no tutor's study of ART1 network, in the present embodiment, training for neural network in the system, adopt no tutor's training and have the tutor to train and combine, (three classes are totally 99 characters promptly to finish all categories in system, see Appendix) the sampling learning tasks after, Adjustment System accuracy of identification controlling value, close the newly-increased classification function of ART1 network, proceed the high precision recognition training, when system prompt has new classification to produce,, continually strengthen accuracy of identification and adaptability by artificial selection and classification.
In concrete the enforcement, be after system finishes the sampling learning tasks of all categories, Adjustment System accuracy of identification control thresholding ρ value to 0.9, close the newly-increased classification function of ART1 network, proceed the high precision recognition training, when system prompt has new classification to produce, by artificial selection and classification, the recognition network weight is continued to optimize in repetition training; Under the situation that system progressively adapts to 2000 samples, progressively reduce thresholding ρ value, use 4000 new samples instead and constantly have acclimatization training; After training is finished, close training function, with 1000 new samples system is tested, as reach the expectation accuracy of identification, then the expression training completes successfully.This training method has effectively overcome the intrinsic drawback of ART1.
In concrete the enforcement,, adopt compound method of identification in order further to improve accuracy of identification, double identification, the then output that comes to the same thing, then identification again of difference as a result, in the possible identification processing time,, discern output until identical the finishing of front and back two times result through repeatedly identification continuously.
Self-elevating platform ART (ART:Adaptive resonance theory) is that American scholar Carpenter proposed in 1976, with Crossbty the ART network is proposed again subsequently, and develop into the self-adaptation concussion ART1 network of handling binary mode and the ART2 type network of handling the continuous analog signal, developed the modified ART network based on ART thought such as ART3, FART, ARTMAP successively again.Basic ART network is formed by comparing layer C, identification layer R and control signal G1, G2, thresholding ρ as shown in Figure 2.The connection weight vector of C and R is included in respectively among C and the R.
Referring to accompanying drawing 3, neural network in the present embodiment, the ART1 network of three standards that employing is had one's own work to do comprises Chinese Character Recognition network A RT1-1, English alphabet recognition network ART1-2 and digit recognition network A RT1-3, and three standard networks are discerned input vector simultaneously; Since three neural networks the training stage respectively the height cluster in recognition objective object range separately, to one's name the input object of recognition objective can " not interested " in other, the similarity of corresponding output recognition result generally can be too not high, handle through the identification of three neural networks, the individuality that three neural networks are exported the class that similarity is the highest is separately simultaneously respectively sent in the thresholding filtration module; In the thresholding filtration module, licence plate character alignment attribute characteristics by the number plate template and identification character and similarity value are carried out thresholding filtering, recognition system accurately selects a unique recognition result character to put into the number plate buffer zone that number plate is accurately discerned filtration module, and recognition system is also judged the number plate type according to first character value and the number plate template of each number plate here; Accurately discern in the filtration module at number plate, system at first finishes the assembling of a complete number plate, next finishes the comparison of double number plate recognition result, when two complete number plate recognition results are consistent, system adds the number plate color value that obtains in the number plate image processing section, export as the system identification result, this module is also undertaken number plate output inhibit feature simultaneously, after a recognition result output, if next wait to export recognition result and last one identical, then system does not allow output, has guaranteed that like this a number plate of vehicle has only a recognition result output.The precision control signal of each ART1 network is mainly undertaken the precision in each ARTI network learning and training stage and is adjusted task, the learning efficiency of control system and the cluster of system in figure three.
For adopting compound method of identification, when car speed is 150 kilometers of speed per hours, about 41 meters of per second speed, the video camera picture depth of field of effectively finding a view is about 5 meters, about 400 milliseconds of maximum effectively processing time of computing machine, the computing machine single frames picture identification processing time needs 125 milliseconds approximately, can guarantee so repeatedly to discern continuously, until the correct recognition result of output.Usually general speed of a motor vehicle speed per hour is no more than 80 kilometers in the urban district, and therefore, this method is more reliable.
With " Anhui A81700 " number plate is example, after certain this License Plate image is hunted down, pass through system handles, to first Chinese Character Recognition, three ART1 export " Anhui ", " K ", " 4 " respectively, and the similarity of three characters is respectively that " Anhui " is 85%, " K " is 82%, " 4 " are 75%, three by the number plate range analysis, because first is Chinese character, therefore the similarity of " Anhui " word identification simultaneously is judged as " Anhui " and output greater than 75% (system preestablishes); Second of corresponding number plate is letter " A ", three ART1 export " penta ", " A ", " 4 " respectively, the similarity of three characters is respectively that " penta " is 82%, " A " is 94%, " 4 " are 81%, three by the number plate range analysis, because second is English alphabet, therefore the similarity of " A " word identification simultaneously is judged as " A " and output greater than 75%; For the 3rd output, system exports " Shan " 78%, " B " 80%, " 8 " 96% respectively, from the number plate range analysis as can be known, this position may be letter or number, the similarity that then compares ART1-2 and ART1-3 output, the unique output of the conduct that similarity is big, " 8 " output is selected by system; The 4th similar to the 3rd identification, and system exports " fourth " 58%, " T " 60%, " 1 " 100% respectively, and system selects " 1 " output automatically; To the 5th, system exports " fourth " 83%, " T " 91%, " 7 " 92% respectively, and " 7 " output is selected by system; To the 6th, system exports " Anhui " 79%, " M " 79%, " 0 " 77% respectively, can only be numeral, system's output " 0 "; In like manner to the 7th, system exports " disappearing " 68%, " R " 74%, " 0 " 77% respectively, and system selects " 0 " as unique output; Each License Plate is because reasons such as far and near size dimension difference, light variation, inclinations, it is different exporting the result respectively during each identification, even one pair of identical number plate, owing to reasons such as positions, the result of three ART network outputs is also inequality during each identification, but because system adopts complex art to handle, can obtain unique accurate recognition result output, its basis and the crucial accurate identification output that also is based on the ART neural network.When a complete number plate analysis finished, record result, and same vehicle is carried out second take turns identification, and with result and result's comparison last time as identical, exported as recognition result after then adding the number plate color attribute; As difference, abandon recognition result last time, write down current results automatically, turn to next identifying.When the speed of a motor vehicle is 140 kilometers, can finish 3-4 identifying, so can discern exactly calmly.Alphabetical, digital discrimination is very high, can reach 100% substantially.Identification for Chinese character, owing to reasons such as stained, the peeling paint of number plate, light variation, inclinations, may cause its similarity not reach setting value, during number plate identification, automatically carry out fault-tolerant processing by the ground Domain Properties characteristics of number plate, directly the Chinese character with the location number plate replaces; For example to the identification of " Anhui " word, certain recognition result is " Jiangxi ", and similarity is 69%, be lower than 75% requirement of default, use the region to be the area, Anhui Province as system, then system replaces output from employing local License Plate " Anhui " word, as much as possible improves accuracy of identification.
Appendix:
1, car plate Chinese character: totally 64
Capital, Tianjin, Ji, Shanxi, illiteracy, the Liao Dynasty, Ji, black, Shanghai, Soviet Union, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, Henan, Hubei Province, Hunan,
Guangdong, osmanthus, fine jade, river, expensive, cloud, Tibetan, Shan, sweet, blue or green, peaceful, new, Chongqing, Hong Kong, Macao, Taiwan, first, second,
The third, fourth, penta, oneself, heptan, suffering, the ninth of the ten Heavenly Stems, the last of the ten Heavenly stems, son, ugly, third of the twelve Earthly Branches, the fourth of the twelve Earthly Branches, occasion,, the noon, not, Shen, the tenth of the twelve Earthly Branches,
The eleventh of the twelve Earthly Branches, the last of the twelve Earthly Branches, police,, spy, make, disappear, limit, logical, WJ
2, English character: totally 25
A、B、C、D、E、F、G、H、J、K、L、M、N、O、P、Q、R、S、T、U、V、
W、X、Y
3, arabic numeric characters: totally 10: 1,2,3,4,5,6,7,8,9,0
Claims (4)
1. identification method of mobile number plate based on three-channel parallel artificial nerve network is characterized in that identification step is as follows:
1), gather the processing of License Plate video image and signal acquisition:
A, employing video camera to the collection of moving vehicle video image, adopt the sport video triggering mode to trigger automatically and carry out video capture.
B, frequently signal acquisition carries out after the Character segmentation input signal as neural network.
2), described neural network adopts the self-adaptation concussion neural network (ART1) of three standards of having one's own work to do, be Chinese Character Recognition network (ART1-1), English alphabet recognition network (ART1-2) and digit recognition network (ART1-3), three standard networks are discerned input signal simultaneously, and its operation steps is:
Three parallel neural networks respectively trained, make recognition system reach required accuracy of identification and adaptability a., earlier.
B, input signal is discerned:
1., from described three recognition networks, to exporting three the highest classification individualities of similarity respectively after the input signal identification, character alignment attribute characteristics by predefined License Plate template, individual to described three classifications output, carry out uniqueness according to number plate character field value attribute and select, guarantee that the result of each character bit output is that most probable and similarity are the highest;
2., when similarity does not reach setting value, handle by fault-tolerant mode and to export;
3., when a complete License Plate through filtering, after the double accurate recognition result unanimity, add the number plate color attribute and export as recognition result.
2, identification method of mobile number plate based on three-channel parallel artificial nerve network according to claim 1, the automatic triggering that it is characterized in that described sport video triggering mode is, when moving vehicle enters the default virtual detection coil of screen, to cause the variation of video image, recognition system adopts the License Plate graphics template that picture is scanned fast, when finding that possibility number plate zone exists, the automatic mark of system, after the high-velocity scanning of finishing full frame, again analysis and judgement is carried out in a plurality of zones of being found, find out real number plate and cut apart and handle, can finish like this there is the seizure of a plurality of License Plates in single-image and improves capture velocity.
3, the identification method of mobile number plate based on three-channel parallel artificial nerve network according to claim 1, it is characterized in that the training of described neural network, be to adopt no tutor's training and have the tutor to train to combine, promptly after the three-channel parallel neural network is finished the sampling learning tasks of all categories respectively, Adjustment System accuracy of identification controlling value, close the newly-increased classification function of ART1 network, proceed the high precision recognition training, when system prompt has new classification to produce, by artificial selection and classification, continually strengthen accuracy of identification and adaptability.
4, the identification method of mobile number plate based on three-channel parallel artificial nerve network according to claim 1, the identification that is described neural network is to adopt compound method of identification, double identification, then output comes to the same thing, then identification again of difference as a result, in the possible identification processing time,, discern output until identical the finishing of front and back two times result through repeatedly identification continuously.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100245936A CN100357988C (en) | 2005-03-24 | 2005-03-24 | Identification method of mobile number plate based on three-channel parallel artificial nerve network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100245936A CN100357988C (en) | 2005-03-24 | 2005-03-24 | Identification method of mobile number plate based on three-channel parallel artificial nerve network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1694130A true CN1694130A (en) | 2005-11-09 |
| CN100357988C CN100357988C (en) | 2007-12-26 |
Family
ID=35353082
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005100245936A Expired - Fee Related CN100357988C (en) | 2005-03-24 | 2005-03-24 | Identification method of mobile number plate based on three-channel parallel artificial nerve network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100357988C (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100394437C (en) * | 2006-02-28 | 2008-06-11 | 浙江工业大学 | Dynamic character discriminating method of digital instrument based on BP nerve network |
| CN102298811A (en) * | 2010-06-25 | 2011-12-28 | 无锡中星微电子有限公司 | Handheld terminal |
| CN101944174B (en) * | 2009-07-08 | 2012-06-27 | 西安电子科技大学 | Identification method of characters of licence plate |
| CN101719272B (en) * | 2009-11-26 | 2012-07-04 | 上海大学 | Three-dimensional image segmentation method based on three-dimensional improved pulse coupled neural network |
| CN102610119A (en) * | 2012-03-22 | 2012-07-25 | 广州杰赛科技股份有限公司 | Reverse car locating method and reverse car locating system |
| CN104036271A (en) * | 2014-06-11 | 2014-09-10 | 新达通科技股份有限公司 | Method and device for identifying crown word number of paper money and character, and ATM (Automatic Teller Machine) |
| CN104464302A (en) * | 2014-12-22 | 2015-03-25 | 南京中兴力维软件有限公司 | License plate recognition and intelligent error correction method and system |
| CN105740906A (en) * | 2016-01-29 | 2016-07-06 | 中国科学院重庆绿色智能技术研究院 | Depth learning based vehicle multi-attribute federation analysis method |
| WO2016197381A1 (en) * | 2015-06-12 | 2016-12-15 | Sensetime Group Limited | Methods and apparatus for recognizing text in an image |
| CN106326897A (en) * | 2016-08-24 | 2017-01-11 | 王小兰 | License plate number recognition method based on data model |
| CN106446892A (en) * | 2016-09-13 | 2017-02-22 | 东软集团股份有限公司 | Vehicle license plate recognizing method, apparatus and charging pile |
| CN107667382A (en) * | 2015-06-26 | 2018-02-06 | 莱克斯真株式会社 | Vehicle number code recognition device and its method |
| CN107703146A (en) * | 2017-09-30 | 2018-02-16 | 北京得华机器人技术研究院有限公司 | A kind of auto-parts vision detection system and method |
| CN107944450A (en) * | 2017-11-16 | 2018-04-20 | 深圳市华尊科技股份有限公司 | A kind of licence plate recognition method and device |
| CN108052866A (en) * | 2017-11-17 | 2018-05-18 | 克立司帝控制系统（上海）有限公司 | Car license recognition learning method and system based on artificial neural network |
| CN108256541A (en) * | 2016-12-29 | 2018-07-06 | 杭州海康威视数字技术股份有限公司 | Car plate index identification method and device |
| CN109389609A (en) * | 2018-10-24 | 2019-02-26 | 西安电子科技大学 | Interaction self feed back infrared target detection method based on FART neural network |
| CN109697857A (en) * | 2019-01-18 | 2019-04-30 | 合肥米佑信息技术有限公司 | Intelligent traffic control system based on image recognition and neural network algorithm |
| CN109840523A (en) * | 2018-12-29 | 2019-06-04 | 南京睿速轨道交通科技有限公司 | A kind of municipal rail train Train number recognition algorithm based on image procossing |
| CN109919969A (en) * | 2019-01-22 | 2019-06-21 | 广东工业大学 | A method of realizing that visual movement controls using depth convolutional neural networks |
| CN110321585A (en) * | 2019-04-09 | 2019-10-11 | 国网山西省电力公司电力科学研究院 | Based on GA-BP neural network switchgear method for detecting insulation defect and system |
| CN113326809A (en) * | 2021-06-30 | 2021-08-31 | 重庆大学 | Off-line signature identification method and system based on three-channel neural network |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| FR2617309B1 (en) * | 1987-06-29 | 1993-07-16 | Cga Hbs | SYSTEM FOR THE AUTOMATIC READING OF IDENTIFICATION DATA, APPOINTED ON A VEHICLE |
| CN2181048Y (en) * | 1993-10-19 | 1994-10-26 | 沈阳市公安交通工程技术研究所 | Automatic identifying and recording apparatus fornumber board of motive vehicle |
| US6026177A (en) * | 1995-08-29 | 2000-02-15 | The Hong Kong University Of Science & Technology | Method for identifying a sequence of alphanumeric characters |
| CN1420041A (en) * | 2001-11-21 | 2003-05-28 | 北京汉王科技有限公司 | Embedded integrative vehicle licensing plate distinguishing apparatus |
| US20030133594A1 (en) * | 2002-01-11 | 2003-07-17 | Sefton Alan K. | Apparatus for method of recording a location of an object having an externally viewable identifier |
-
2005
- 2005-03-24 CN CNB2005100245936A patent/CN100357988C/en not_active Expired - Fee Related
Cited By (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100394437C (en) * | 2006-02-28 | 2008-06-11 | 浙江工业大学 | Dynamic character discriminating method of digital instrument based on BP nerve network |
| CN101944174B (en) * | 2009-07-08 | 2012-06-27 | 西安电子科技大学 | Identification method of characters of licence plate |
| CN101719272B (en) * | 2009-11-26 | 2012-07-04 | 上海大学 | Three-dimensional image segmentation method based on three-dimensional improved pulse coupled neural network |
| CN102298811A (en) * | 2010-06-25 | 2011-12-28 | 无锡中星微电子有限公司 | Handheld terminal |
| CN102298811B (en) * | 2010-06-25 | 2015-06-24 | 无锡中星微电子有限公司 | Handheld terminal |
| CN102610119A (en) * | 2012-03-22 | 2012-07-25 | 广州杰赛科技股份有限公司 | Reverse car locating method and reverse car locating system |
| CN102610119B (en) * | 2012-03-22 | 2014-04-02 | 广州杰赛科技股份有限公司 | Reverse car locating method and reverse car locating system |
| CN104036271B (en) * | 2014-06-11 | 2017-08-25 | 新达通科技股份有限公司 | The recognition methods of a kind of character and paper money number and device and ATM |
| CN104036271A (en) * | 2014-06-11 | 2014-09-10 | 新达通科技股份有限公司 | Method and device for identifying crown word number of paper money and character, and ATM (Automatic Teller Machine) |
| CN104464302A (en) * | 2014-12-22 | 2015-03-25 | 南京中兴力维软件有限公司 | License plate recognition and intelligent error correction method and system |
| CN107636691A (en) * | 2015-06-12 | 2018-01-26 | 商汤集团有限公司 | Method and apparatus for identifying the text in image |
| WO2016197381A1 (en) * | 2015-06-12 | 2016-12-15 | Sensetime Group Limited | Methods and apparatus for recognizing text in an image |
| CN107667382A (en) * | 2015-06-26 | 2018-02-06 | 莱克斯真株式会社 | Vehicle number code recognition device and its method |
| CN105740906A (en) * | 2016-01-29 | 2016-07-06 | 中国科学院重庆绿色智能技术研究院 | Depth learning based vehicle multi-attribute federation analysis method |
| CN105740906B (en) * | 2016-01-29 | 2019-04-02 | 中国科学院重庆绿色智能技术研究院 | A kind of more attribute conjoint analysis methods of vehicle based on deep learning |
| CN106326897A (en) * | 2016-08-24 | 2017-01-11 | 王小兰 | License plate number recognition method based on data model |
| CN106446892A (en) * | 2016-09-13 | 2017-02-22 | 东软集团股份有限公司 | Vehicle license plate recognizing method, apparatus and charging pile |
| CN108256541A (en) * | 2016-12-29 | 2018-07-06 | 杭州海康威视数字技术股份有限公司 | Car plate index identification method and device |
| CN107703146A (en) * | 2017-09-30 | 2018-02-16 | 北京得华机器人技术研究院有限公司 | A kind of auto-parts vision detection system and method |
| CN107944450A (en) * | 2017-11-16 | 2018-04-20 | 深圳市华尊科技股份有限公司 | A kind of licence plate recognition method and device |
| CN108052866A (en) * | 2017-11-17 | 2018-05-18 | 克立司帝控制系统（上海）有限公司 | Car license recognition learning method and system based on artificial neural network |
| CN109389609A (en) * | 2018-10-24 | 2019-02-26 | 西安电子科技大学 | Interaction self feed back infrared target detection method based on FART neural network |
| CN109389609B (en) * | 2018-10-24 | 2021-11-23 | 西安电子科技大学 | Interactive self-feedback infrared target detection method based on FART neural network |
| CN109840523A (en) * | 2018-12-29 | 2019-06-04 | 南京睿速轨道交通科技有限公司 | A kind of municipal rail train Train number recognition algorithm based on image procossing |
| CN109840523B (en) * | 2018-12-29 | 2023-08-22 | 南京睿速轨道交通科技有限公司 | Urban rail train number identification method based on image processing |
| CN109697857A (en) * | 2019-01-18 | 2019-04-30 | 合肥米佑信息技术有限公司 | Intelligent traffic control system based on image recognition and neural network algorithm |
| CN109919969A (en) * | 2019-01-22 | 2019-06-21 | 广东工业大学 | A method of realizing that visual movement controls using depth convolutional neural networks |
| CN109919969B (en) * | 2019-01-22 | 2022-11-22 | 广东工业大学 | Method for realizing visual motion control by utilizing deep convolutional neural network |
| CN110321585A (en) * | 2019-04-09 | 2019-10-11 | 国网山西省电力公司电力科学研究院 | Based on GA-BP neural network switchgear method for detecting insulation defect and system |
| CN113326809A (en) * | 2021-06-30 | 2021-08-31 | 重庆大学 | Off-line signature identification method and system based on three-channel neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100357988C (en) | 2007-12-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100357988C (en) | Identification method of mobile number plate based on three-channel parallel artificial nerve network | |
| CN110969160B (en) | License plate image correction and recognition method and system based on deep learning | |
| CN103761531B (en) | The sparse coding license plate character recognition method of Shape-based interpolation contour feature | |
| CN102163284B (en) | Chinese environment-oriented complex scene text positioning method | |
| CN107316001A (en) | Small and intensive method for traffic sign detection in a kind of automatic Pilot scene | |
| CN106886778B (en) | License plate character segmentation and recognition method in monitoring scene | |
| CN105809121A (en) | Multi-characteristic synergic traffic sign detection and identification method | |
| CN112329542B (en) | SAR image ship target detection method based on feature refined network model | |
| CN108830254B (en) | Fine-grained vehicle type detection and identification method based on data balance strategy and intensive attention network | |
| CN106529532A (en) | License plate identification system based on integral feature channels and gray projection | |
| CN106709530A (en) | License plate recognition method based on video | |
| CN101980242A (en) | Human face discrimination method and system and public safety system | |
| CN109886147A (en) | A kind of more attribute detection methods of vehicle based on the study of single network multiple-task | |
| Wang | License plate recognition via convolutional neural networks | |
| Zhang et al. | License plate localization in unconstrained scenes using a two-stage CNN-RNN | |
| CN111967313A (en) | Unmanned aerial vehicle image annotation method assisted by deep learning target detection algorithm | |
| CN111965636A (en) | Night target detection method based on millimeter wave radar and vision fusion | |
| CN106919939B (en) | A kind of traffic signboard tracks and identifies method and system | |
| CN105930831A (en) | License plate intelligent identification method | |
| CN106295532A (en) | A kind of human motion recognition method in video image | |
| CN107590500A (en) | A kind of color recognizing for vehicle id method and device based on color projection classification | |
| CN107704833A (en) | A kind of front vehicles detection and tracking based on machine learning | |
| CN109165602A (en) | A kind of black smoke vehicle detection method based on video analysis | |
| CN109635799A (en) | A kind of recognition methods of gas meter, flow meter wheel digital | |
| CN113610024B (en) | Multi-strategy deep learning remote sensing image small target detection method |
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 | ||
| ASS | Succession or assignment of patent right |
Owner name: HEFEI VOCATION TECHNOLOGY CO., LTD. Free format text: FORMER OWNER: SHANGHAI UNIVERSITY Effective date: 20110908 Free format text: FORMER OWNER: HEFEI VOCATION TECHNOLOGY CO., LTD. |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| COR | Change of bibliographic data |
Free format text: CORRECT: ADDRESS; FROM: 200072 ZHABEI, SHANGHAI TO: 230088 HEFEI, ANHUI PROVINCE |
|
| TR01 | Transfer of patent right |
Effective date of registration: 20110908 Address after: 230088 Anhui high tech Zone, Mount Huangshan Road, No. 605, the people of the center of the room, room 508 Patentee after: Hefei Wen Kang Technology Co., Ltd. Address before: 200072 Zhabei District Yanchang Road, Shanghai, No. 149 Co-patentee before: HEFEI VOCATION TECHNOLOGY CO., LTD. Patentee before: Shanghai University |
|
| CI01 | Correction of invention patent gazette |
Correction item: Patentee Correct: HEFEI VOCATION TECHNOLOGY CO., LTD.|230088 Anhui high tech Zone, Mount Huangshan Road, No. 605, the people of the center of the room, room 508 False: Hefei Wen Kang Technology Co., Ltd.|230088 Anhui high tech Zone, Mount Huangshan Road, No. 605, the people of the center of the room, room 508 Number: 42 Volume: 27 |
|
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: Identification method of mobile number plate based on three-channel parallel artificial nerve network Effective date of registration: 20120823 Granted publication date: 20071226 Pledgee: Hefei high tech Company limited by guarantee Pledgor: HEFEI VOCATION TECHNOLOGY CO., LTD. Registration number: 2012990000464 |
|
| PC01 | Cancellation of the registration of the contract for pledge of patent right |
Date of cancellation: 20131112 Granted publication date: 20071226 Pledgee: Hefei high tech Company limited by guarantee Pledgor: HEFEI VOCATION TECHNOLOGY CO., LTD. Registration number: 2012990000464 |
|
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: Identification method of mobile number plate based on three-channel parallel artificial nerve network Effective date of registration: 20131114 Granted publication date: 20071226 Pledgee: Hefei high tech Company limited by guarantee Pledgor: HEFEI VOCATION TECHNOLOGY CO., LTD. Registration number: 2013990000845 |
|
| PLDC | Enforcement, change and cancellation of contracts on pledge of patent right or utility model | ||
| C56 | Change in the name or address of the patentee |
Owner name: ANHUI VOCAT ION TECHNOLOGY CO., LTD. Free format text: FORMER NAME: HEFEI VOCATION TECHNOLOGY CO., LTD. |
|
| CP01 | Change in the name or title of a patent holder |
Address after: 230088 Anhui high tech Zone, Mount Huangshan Road, No. 605, the people of the center of the room, room 508 Patentee after: ANHUI VOCATION TECHNOLOGY CO., LTD. Address before: 230088 Anhui high tech Zone, Mount Huangshan Road, No. 605, the people of the center of the room, room 508 Patentee before: HEFEI VOCATION TECHNOLOGY CO., LTD. |
|
| DD01 | Delivery of document by public notice |
Addressee: Gu Juanjuan Document name: Notification of Passing Examination on Formalities |
|
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20071226 Termination date: 20200324 |