CN105787437A - Vehicle brand type identification method based on cascading integrated classifier - Google Patents
Vehicle brand type identification method based on cascading integrated classifier Download PDFInfo
- Publication number
- CN105787437A CN105787437A CN201610077560.6A CN201610077560A CN105787437A CN 105787437 A CN105787437 A CN 105787437A CN 201610077560 A CN201610077560 A CN 201610077560A CN 105787437 A CN105787437 A CN 105787437A
- Authority
- CN
- China
- Prior art keywords
- classifier
- vehicle
- integrated classifier
- vehicle brand
- adopt
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/285—Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Abstract
The invention discloses a method for identifying vehicle brand types on the basis of a cascading integrated classifier, and provides a vehicle brand type identification method based on a cascading integrated classifier. A vehicle face image is positioned and segmented; vehicle face image features are extracted respectively by use of a gradient orientated histogram and Contourlet transformation, and a first-stage integrated classifier for vehicle brand type identification is constructed; and a multilayer nerve network is taken as a sub-classifier of a second-stage integrated classifier, and a vehicle brand type which a vehicle belongs to is determined through the first-stage integrated classifier and the second-stage integrated classifier. Through such steps, the precision and the reliability of an identification system are greatly enhanced, and an important parameter basis is provided for effectively solving such traffic problems of vehicle plate duplication, traffic offences and the like.
Description
Technical field
Patent of the present invention relates to intelligent transportation research field, especially the research of city vehicle monitoring and management.
Background technology
Along with the development of " safe city ", the identification of vehicle brand type has become one of study hotspot content in field of traffic control.Vehicle brand type identification is particularly important in channel control system (ACS), if parking lot, building and restricted area are by identifying that false (set) the board car of vehicle brand type restriction enters.The research of current vehicle identification is confined to the licence plate unique identities feature as vehicle mostly, for the problem that the vehicle of two different brands types has identical licence plate (namely stealing board), Vehicle License Plate Recognition System cannot check of genuineness, therefore, vehicle brand categorizing system is in field extensive application such as urban transportation monitoring, emergency command and accident detections.
Vehicle brand type identification technology based on video image is with a wide range of applications, and the identification key technology realizing vehicle brand type is the extraction of vehicle characteristics and the structure of grader is chosen.At present, Petrivic proposes a kind of Gradient Features according to image and characterizes type of vehicle, and adopt minimum distance classification method to carry out vehicle type recognition, although some certain types of vehicle can be identified by experimental result automatically, but overall accuracy rate need to improve.
Therefore, it is necessary to a kind of new technical scheme is to solve the problems referred to above
Summary of the invention
Patent of the present invention problem to be solved is automatically to identify vehicle brand type, for the application at false (set) board of vehicle and channel control system.
For solving above-mentioned technical problem, patent of the present invention adopts the following technical scheme that
A kind of vehicle brand kind identification method based on cascade integrated classifier, including step:
1) vehicle pictures of electronic police bayonet socket camera shooting is adopted the license plate area of traversal matching algorithm search vehicle, and the proportionate relationship according to car face region Yu license plate area positions and partition cart face image；
2) it is respectively adopted gradient orientation histogram and contourlet transformation extracts car face characteristics of image, tectonic association characteristic vector, and adopt PCA that the assemblage characteristic vector extracted is carried out dimension-reduction treatment；
3) based on Bayes classifier, k-Nearest Neighbor Classifier, multilayer neural network grader and support vector machine classifier, the first order integrated classifier of structure vehicle brand type identification, and adopt the vector of the assemblage characteristic after dimension-reduction treatment that first order integrated classifier is trained, adopt 4 kinds of Decision Classfication mechanism that each input sample is voted, make it have " rejection " function, separate " rejection " sample；
4) adopt multilayer neural network as the sub-classifier of second level integrated classifier, and adopt the second level integrated classifier rotating forest structure vehicle brand identification, " rejection " sample of the first order is carried out Classification and Identification again；
5) adopt the vector of the assemblage characteristic after dimension-reduction treatment that the cascade integrated classifier constructed is trained, test specimens when not having " rejection " this judge vehicle brand type belonging to it according to its confidence value, and for grader, there is the test sample in " rejection " option situation, then adopt second level integrated classifier to judge vehicle brand type belonging to it.
Compared with prior art, the technical program identifies vehicle brand type with a kind of method based on cascade integrated classifier, considerably increase precision and the reliability of identification system, provide important parameter foundation for efficiently solving the traffic problems such as vehicle fake-license, traffic accident.
Accompanying drawing explanation
Fig. 1 is the cascade integrated classifier schematic diagram in the present invention；
Fig. 2 is the schematic diagram of the first order integrated classifier in the present invention；
Fig. 3 is the schematic diagram of the second level integrated classifier in the present invention.
Detailed description of the invention
Below in conjunction with the drawings and specific embodiments, it is further elucidated with the present invention, it should be understood that these embodiments are merely to illustrate the present invention rather than restriction the scope of the present invention, after having read the present invention, the amendment of the various equivalent form of values of the present invention is all fallen within the application claims limited range by those skilled in the art
A kind of vehicle brand kind identification method based on cascade integrated classifier disclosed by the invention, comprises the following steps:
1) vehicle pictures of electronic police bayonet socket camera shooting is adopted the license plate area of traversal matching algorithm search vehicle, and the proportionate relationship according to car face region Yu license plate area positions and partition cart face image；
2) it is respectively adopted gradient orientation histogram (HOG) and contourlet transformation extracts car face characteristics of image, tectonic association characteristic vector, and adopt PCA (PCA) that the assemblage characteristic vector extracted is carried out dimension-reduction treatment；
3) based on Bayes classifier (Bayes), k-Nearest Neighbor Classifier (kNN), multilayer neural network grader (MLP) and support vector machine classifier (SVM), the first order integrated classifier of structure vehicle brand type identification, and adopt the vector of the assemblage characteristic after dimension-reduction treatment that first order integrated classifier is trained, adopt 4 kinds of Decision Classfication mechanism that each input sample is voted, make it have " rejection " function, separate " rejection " sample；
4) adopt multilayer neural network (MLP) as the sub-classifier of second level integrated classifier, and adopt rotation forest (RandomForest) to construct the second level integrated classifier of vehicle brand identification, " rejection " sample of the first order is carried out Classification and Identification again；
5) adopt the vector of the assemblage characteristic after dimension-reduction treatment that the cascade integrated classifier constructed is trained, test specimens when not having " rejection " this judge vehicle brand type belonging to it according to its confidence value, and for grader, there is the test sample in " rejection " option situation, then adopt second level integrated classifier to judge vehicle brand type belonging to it.
Concrete, described step 1) in, the vehicle pictures of electronic police bayonet socket camera shooting is adopted the license plate area traveling through matching algorithm search vehicle, and the proportionate relationship according to car face region Yu license plate area positions and partition cart face image:
First, positioning licence plate symmetrical centre, and make car plate be sized to W × H, traversal coupling is carried out in Plate searching region, add up each object pixel number being sized in W × H window, being license plate area when object pixel number exceedes a certain threshold value by correspondence window indicia, otherwise get rid of this window, number of pixels threshold value meets claimed below:
Wherein,For object pixel numbers all in region of search, coefficient ρ takes 1.2, the size of decision threshold t.Owing to there is fixed relationship between face picture size and license plate image before vehicle, making car plate width is Wp, with width for 2.6Wp, height is 1.1WpRectangular area extract face region before vehicle.
Described step 2) in, it is respectively adopted gradient orientation histogram (HOG) and contourlet transformation extracts car face characteristics of image, tectonic association characteristic vector, and adopt PCA (PCA) that the assemblage characteristic vector extracted is carried out dimension-reduction treatment:
Gradient orientation histogram (HOG) feature extraction is the statistical value of topography's Gradient direction information.First car face image is divided into little connected region, i.e. cell factory (Cells), then calculate according to the gradient direction of pixel each in Cells and amplitude and obtain its gradient orientation histogram, finally these set of histograms constitutive characteristic describer altogether.In order to improve performance, the gradient orientation histogram of these Cells is normalized in the bigger interval (Block) of image, namely each rectangular histogram density in interval is first calculated, then according to density, each cell factory in interval is normalized, the impact of illumination and shade can be eliminated by normalization.
Contourlet transformation is the image representing method of a kind of multiresolution realized in conjunction with Laplacian pyramid decomposition (LP) and directional filter banks (DFB), multi-direction and local, approach image with " strip " based structures with dimensional variation length-width ratio, there is directivity and anisotropy.In car face image characteristics extraction in this patent, feature vector dimension can be caused too high directly as input feature vector sub-band images, but the direct average using each subband and variance can neglect again the detailed information that comparison is many as input feature vector, therefore, this patent proposes the average of each row of the image after adopting all subband of each layer to merge and variance as feature.
HOG feature according to the car face image extracted respectively and Contourlet feature, adopt paralleling tactic to build the assemblage characteristic vector of car face image, and adopt PCA (PCA) that the assemblage characteristic vector in car face region is carried out dimension-reduction treatment.PCA is the process that data space is mapped to lower-dimensional subspace by orthogonal transformation, by trying to achieve the projection matrix of a low-dimensional, is multiplied by this projection matrix by the feature of higher-dimension, just the dimension of the high dimensional feature of vehicle image can be dropped to the dimension specified.
Step 3) in, based on Bayes classifier (Bayes), k-Nearest Neighbor Classifier (kNN), multilayer neural network grader (MLP) and support vector machine classifier (SVM), the first order integrated classifier of structure vehicle brand type identification, and adopt the vector of the assemblage characteristic after dimension-reduction treatment that first order integrated classifier is trained, adopt 4 kinds of Decision Classfication mechanism that each input sample is voted, make it have " rejection " function, separate " rejection " sample.
The cascade integrated classifier scheme that this patent adopts is as shown in Figure 1, including two-stage integrated classifier, the input test sample of the integrated classifier of the second level is the sample being rejected identification in first order integrated classifier, and first order integrated classifier scheme is as shown in Figure 2.In first order integrated classifier, it is respectively adopted assemblage characteristic vector training Bayes classifier (Bayes), k-Nearest Neighbor Classifier (kNN), multilayer neural network (MLP) grader and support vector machine (SVM) grader that adopt the car face after dimension-reduction treatment, the final type of vehicle brand is to be chosen in a vote by the sub-classifier in first order integrated classifier, when the rarest 3 sub-classifiers agree to a certain classification results, just give this value by this car face characteristic vector；In other cases, for certain confidence level less than 3 input sample sub-classifier will abandon the identification to it so that it is there is " rejection " function, separate " rejection " sample.
Step 4) in, adopt multilayer neural network (MLP) as the sub-classifier of second level integrated classifier, and adopt rotation forest (RandomForest) to construct the second level integrated classifier of vehicle brand identification, " rejection " sample of the first order is carried out Classification and Identification again.
Adopting multilayer neural network (MLP) as the sub-classifier of second level integrated classifier, and adopt rotation forest (RandomForest) to construct the second level grader of vehicle brand type identification, the program is as shown in Figure 3.The number making the MLP sub-classifier in the integrated classifier of the second level is M2, when the rarest T sub-classifier agrees to a certain classification results, give this classification results by input test sample type value.Here T is decision-making value, works as M2When being even number,M2When being odd number,If the prediction of certain sample is not reached an agreement by grader, then still refuse the identification to it.
Step 5) in, adopt the vector of the assemblage characteristic after dimension-reduction treatment that the cascade integrated classifier constructed is trained, test specimens when not having " rejection " this judge vehicle brand type belonging to it according to its confidence value, and for grader, there is the test sample in " rejection " option situation, then adopt second level integrated classifier to judge vehicle brand type belonging to it:
The cascade integrated classifier constructed is trained by the assemblage characteristic vector adopting the car face image after dimension-reduction treatment.Test sample x when not having " rejection ", makes di,j(xRi) for grader DiJudge that it is ωjThe probability of class, then each classification ω distributed to by samplejCredibility be:
According to confidence value, x distributed to the classification corresponding to credibility maximum.Grader being had to the test sample x in " rejection " option situation, categorised decision scheme shown in Fig. 3, each base grader dopes a types value, finally uses the method for majority ballot to determine the vehicle brand type that this sample should belong to.
Below with a concrete example to the technical program explanation for example:
The first step: the car plate symmetrical centre of the vehicle pictures of positioning electronic police's bayonet socket camera shooting, and make car plate be sized to W × H；In Plate searching region, carry out traversal coupling, add up each object pixel number being sized in W × H window, be license plate area when object pixel number exceedes threshold value t by correspondence window indicia, otherwise get rid of this window.Owing to there is fixed relationship between face picture size and license plate image before vehicle, with width for 2.6W, face region before vehicle is extracted in the rectangular area that height is 1.1W.
Second step: first, gradient orientation histogram (HOG) is adopted to extract car face characteristics of image, and calculate each rectangular histogram density in interval, according to density, each cell factory in interval is normalized, the impact of illumination and shade can be eliminated by normalization；Secondly, contourlet transformation is adopted to extract car face characteristics of image, the each row average of the image after adopting all subband of each layer to merge and variance are as feature, one input is sized to the image of 256 × 256, three layers subband after contourlet transformation merges image size respectively 16 × 16,64 × 64,128 × 128, then the Contourlet feature vector dimension of this car face image is 2 × (16+64+128)=896；Finally, according to the HOG feature of the car face image extracted respectively and Contourlet feature, adopt paralleling tactic to build the assemblage characteristic vector of car face image, and adopt principal component analysis (PCA) that the assemblage characteristic vector of car face image is carried out dimension-reduction treatment.
3rd step: adopting two-stage cascade integrated classifier scheme, the input test sample of second level integrated classifier is the sample being rejected identification in first order integrated classifier.In first order integrated classifier, it is respectively adopted training Bayes classifier of the assemblage characteristic vector after dimension-reduction treatment (Bayes), k-Nearest Neighbor Classifier (kNN), multilayer neural network (MLP) grader and support vector machine (SVM) grader, the final type of vehicle brand is to be chosen in a vote by the sub-classifier in first order integrated classifier, when the rarest 3 sub-classifiers agree to a certain classification results, just give this value by this characteristic vector；In other cases, for certain confidence level less than 3 input sample sub-classifier will abandon the identification to it so that it is there is " rejection " function, separate " rejection " sample.
4th step: adopt multilayer neural network (MLP) as the sub-classifier of second level integrated classifier, and adopt rotation forest (RandomForest) to construct the second level grader of vehicle brand identification.The number making the MLP sub-classifier in the integrated classifier of the second level is M2, when the rarest t sub-classifier agrees to a certain classification results, give this classification results by input test sample type value.Here t is decision-making value, works as M2When being even number,M2When being odd number,If the prediction of certain sample is not reached an agreement by grader, then still refuse the identification to it.
5th step: adopt the assemblage characteristic vector of the car face after dimension-reduction treatment that the cascade integrated classifier constructed is trained.For there is no the test sample x of " rejection " situation, make di,j(xRi) for grader DiJudge that it is ωjThe probability of class, then each classification ω distributed to by samplejCredibility be ujX (), distributes to the classification corresponding to credibility maximum according to confidence value by x.Grader being had to the test sample x in " rejection " option situation, according to the categorised decision scheme of second level integrated classifier, each base grader dopes a types value, finally uses the method for majority ballot to determine the vehicle brand type that this sample should belong to.
In sum, compared with prior art, the technical program identifies vehicle brand type with a kind of method based on cascade integrated classifier, considerably increases precision and the reliability of identification system, provides important parameter foundation for efficiently solving the traffic problems such as vehicle fake-license, traffic accident.
Claims (7)
1., based on a vehicle brand kind identification method for cascade integrated classifier, it is characterized in that including step:
1) vehicle pictures of electronic police bayonet socket camera shooting is adopted the license plate area of traversal matching algorithm search vehicle, and the proportionate relationship according to car face region Yu license plate area positions and partition cart face image；
2) it is respectively adopted gradient orientation histogram and contourlet transformation extracts car face characteristics of image, tectonic association characteristic vector, and adopt PCA that the assemblage characteristic vector extracted is carried out dimension-reduction treatment；
3) based on Bayes classifier, k-Nearest Neighbor Classifier, multilayer neural network grader and support vector machine classifier, the first order integrated classifier of structure vehicle brand type identification, and adopt the vector of the assemblage characteristic after dimension-reduction treatment that first order integrated classifier is trained, adopt 4 kinds of Decision Classfication mechanism that each input sample is voted, make it have " rejection " function, separate " rejection " sample；
4) adopt multilayer neural network as the sub-classifier of second level integrated classifier, and adopt the second level integrated classifier rotating forest structure vehicle brand identification, " rejection " sample of the first order is carried out Classification and Identification again；
5) adopt the vector of the assemblage characteristic after dimension-reduction treatment that the cascade integrated classifier constructed is trained, test specimens when not having " rejection " this judge vehicle brand type belonging to it according to its confidence value, and for grader, there is the test sample in " rejection " option situation, then adopt second level integrated classifier to judge vehicle brand type belonging to it.
2. the vehicle brand kind identification method based on cascade integrated classifier according to claim 1, it is characterized in that described step 1) in, the vehicle pictures of electronic police bayonet socket camera shooting is adopted the license plate area traveling through matching algorithm search vehicle, and the proportionate relationship according to car face region Yu license plate area positions and partition cart face image；
First, positioning licence plate symmetrical centre, and make car plate be sized to W × H, traversal coupling is carried out in Plate searching region, add up each object pixel number being sized in W × H window, being license plate area when object pixel number exceedes a certain threshold value by correspondence window indicia, otherwise get rid of this window, number of pixels threshold value t meets claimed below:
Wherein,For object pixel numbers all in region of search, coefficient ρ is for making numerical value by oneself, and ρ is in order to determine the size of number of pixels threshold value t；Owing to there is fixed relationship between face picture size and license plate image before vehicle, making car plate width is Wp, with width for 2.6Wp, height is 1.1WpRectangular area extract face region before vehicle.
3. the vehicle brand kind identification method based on cascade integrated classifier according to claim 1, its described step 2) in,
Gradient orientation histogram) feature extraction be the statistical value of topography's Gradient direction information；First car face image is divided into little connected region, i.e. cell factory, then calculate according to the gradient direction of pixel each in cell factory and amplitude and obtain its gradient orientation histogram, finally these set of histograms constitutive characteristic describer altogether；Again the gradient orientation histogram of these cell factory is normalized in the bigger interval of image, namely each rectangular histogram density in interval is first calculated, then according to density, each cell factory in interval is normalized, the impact of illumination and shade can be eliminated by normalization；
Contourlet transformation is the image representing method of a kind of multiresolution realized in conjunction with Laplacian pyramid decomposition and directional filter banks, multi-direction and local, approach image with " strip " based structures with dimensional variation length-width ratio, there is directivity and anisotropy；In car face image characteristics extraction, the average of each row of the image after adopting all subband of each layer to merge and variance are as feature；
HOG feature according to the car face image extracted respectively and Contourlet feature, adopt paralleling tactic to build the assemblage characteristic vector of car face image, and adopt PCA that the assemblage characteristic vector in car face region is carried out dimension-reduction treatment；PCA is the process that data space is mapped to lower-dimensional subspace by orthogonal transformation, by trying to achieve the projection matrix of a low-dimensional, is multiplied by this projection matrix by the feature of higher-dimension, just the dimension of the high dimensional feature of vehicle image can be dropped to the dimension specified.
4. the vehicle brand kind identification method based on cascade integrated classifier according to claim 1, its described step 3) in,
Adopting two-stage cascade integrated classifier scheme, the input test sample of the integrated classifier of the second level is the sample being rejected identification in first order integrated classifier；In first order integrated classifier, it is respectively adopted the training Bayes classifier of the assemblage characteristic vector after dimension-reduction treatment, k-Nearest Neighbor Classifier, multilayer neural network grader and support vector machine classifier, the final type of vehicle brand is to be chosen in a vote by the sub-classifier in first order integrated classifier, when the rarest 3 sub-classifiers agree to a certain classification results, just give this value by this car face characteristic vector；In other cases, for certain confidence level less than 3 input sample sub-classifier will abandon the identification to it so that it is there is " rejection " function, separate " rejection " sample.
5. the vehicle brand kind identification method based on cascade integrated classifier according to claim 2, is characterized in that its described step 4) in,
The number making the MLP sub-classifier in the integrated classifier of the second level is M2, when the rarest T sub-classifier agrees to a certain classification results, give this classification results by input test sample type value；Here T is decision-making value, works as M2When being even number,M2When being odd number,If the prediction of certain sample is not reached an agreement by grader, then still refuse the identification to it.
6. the vehicle brand kind identification method based on cascade integrated classifier according to claim 2, is characterized in that its described step 5) in,
Test sample x when not having " rejection ", makes di,j(xRi) for grader DiJudge that it is ωjThe probability of class, then each classification ω distributed to by samplejCredibility be:
According to confidence value, x distributed to the classification corresponding to credibility maximum；Grader being had to the test sample x in " rejection " option situation, adopt second level integrated classifier categorised decision scheme, each base grader dopes a types value, finally uses the method for majority ballot to determine the vehicle brand type that this sample should belong to.
7. the vehicle brand kind identification method based on cascade integrated classifier according to claim 2, is characterized in that, described ρ takes 1.2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610077560.6A CN105787437B (en) | 2016-02-03 | 2016-02-03 | A kind of vehicle brand kind identification method based on cascade integrated classifier |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610077560.6A CN105787437B (en) | 2016-02-03 | 2016-02-03 | A kind of vehicle brand kind identification method based on cascade integrated classifier |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105787437A true CN105787437A (en) | 2016-07-20 |
| CN105787437B CN105787437B (en) | 2017-04-05 |
Family
ID=56402335
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610077560.6A Active CN105787437B (en) | 2016-02-03 | 2016-02-03 | A kind of vehicle brand kind identification method based on cascade integrated classifier |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105787437B (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106250832A (en) * | 2016-07-25 | 2016-12-21 | 华南理工大学 | A kind of national recognition methods based on integrated convolutional neural networks |
| CN106529542A (en) * | 2016-09-30 | 2017-03-22 | 中国石油天然气股份有限公司 | Indicator diagram identification method and device |
| CN106650612A (en) * | 2016-10-27 | 2017-05-10 | 嘉兴学院 | Road vehicle detection and classification method |
| CN106897695A (en) * | 2017-02-24 | 2017-06-27 | 上海斐讯数据通信技术有限公司 | A kind of image recognizing and processing equipment, system and method |
| CN106971556A (en) * | 2017-05-16 | 2017-07-21 | 中山大学 | The recognition methods again of bayonet vehicle based on dual network structure |
| CN107798281A (en) * | 2016-09-07 | 2018-03-13 | 北京眼神科技有限公司 | A kind of human face in-vivo detection method and device based on LBP features |
| CN108345896A (en) * | 2017-01-23 | 2018-07-31 | 许继集团有限公司 | High-tension apparatus image-recognizing method and device based on variety of components and scale of components |
| CN108388871A (en) * | 2018-02-28 | 2018-08-10 | 中国计量大学 | A kind of vehicle checking method returned based on vehicle body |
| CN108494620A (en) * | 2018-02-28 | 2018-09-04 | 南京邮电大学 | Network service flow feature selecting and sorting technique based on multiple target Adaptive evolvement arithmetic |
| CN108810547A (en) * | 2018-07-03 | 2018-11-13 | 电子科技大学 | A kind of efficient VR video-frequency compression methods based on neural network and PCA-KNN |
| CN109002830A (en) * | 2017-06-06 | 2018-12-14 | 罗伯特·博世有限公司 | The for motor vehicle method and apparatus for object of classification |
| CN109299729A (en) * | 2018-08-24 | 2019-02-01 | 四川大学 | Vehicle checking method and device |
| CN110135318A (en) * | 2019-05-08 | 2019-08-16 | 佳都新太科技股份有限公司 | Cross determination method, apparatus, equipment and the storage medium of vehicle record |
| CN110494902A (en) * | 2017-02-03 | 2019-11-22 | 西门子交通有限责任公司 | For managing the system, apparatus and method of the traffic in geographical location |
| CN110598749A (en) * | 2019-08-13 | 2019-12-20 | 广东数鼎科技有限公司 | Image vehicle recognition implementation system based on convolutional neural network AI algorithm |
| CN110858303A (en) * | 2018-08-22 | 2020-03-03 | 西门子（中国）有限公司 | Method and device for detecting license plate number |
| CN112036421A (en) * | 2019-05-16 | 2020-12-04 | 搜狗（杭州）智能科技有限公司 | Image processing method and device and electronic equipment |
| CN112801233A (en) * | 2021-04-07 | 2021-05-14 | 杭州海康威视数字技术股份有限公司 | Internet of things equipment honeypot system attack classification method, device and equipment |
| CN113537666A (en) * | 2020-04-16 | 2021-10-22 | 马上消费金融股份有限公司 | Evaluation model training method, evaluation and business auditing method, device and equipment |
| US11386680B2 (en) | 2020-03-28 | 2022-07-12 | Wipro Limited | System and method of identifying vehicle brand and model |
| WO2022247573A1 (en) * | 2021-05-26 | 2022-12-01 | 腾讯科技（深圳）有限公司 | Model training method and apparatus, image processing method and apparatus, device, and storage medium |
| CN109002830B (en) * | 2017-06-06 | 2024-04-30 | 罗伯特·博世有限公司 | Method and device for classifying objects for vehicles |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102411710A (en) * | 2011-12-09 | 2012-04-11 | 东南大学 | Vehicle type recognition method based on vehicle face features |
| CN102542257A (en) * | 2011-12-20 | 2012-07-04 | 东南大学 | Driver fatigue level detection method based on video sensor |
| CN103279756A (en) * | 2013-06-13 | 2013-09-04 | 苏州市公安局苏州工业园区分局 | Vehicle detecting analysis system and detecting analysis method thereof based on integrated classifier |
| CN103488973A (en) * | 2013-09-12 | 2014-01-01 | 上海依图网络科技有限公司 | Method and system for recognizing vehicle brand based on image |
| CN103903441A (en) * | 2014-04-04 | 2014-07-02 | 山东省计算中心 | Road traffic state distinguishing method based on semi-supervised learning |
-
2016
- 2016-02-03 CN CN201610077560.6A patent/CN105787437B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102411710A (en) * | 2011-12-09 | 2012-04-11 | 东南大学 | Vehicle type recognition method based on vehicle face features |
| CN102542257A (en) * | 2011-12-20 | 2012-07-04 | 东南大学 | Driver fatigue level detection method based on video sensor |
| CN103279756A (en) * | 2013-06-13 | 2013-09-04 | 苏州市公安局苏州工业园区分局 | Vehicle detecting analysis system and detecting analysis method thereof based on integrated classifier |
| CN103488973A (en) * | 2013-09-12 | 2014-01-01 | 上海依图网络科技有限公司 | Method and system for recognizing vehicle brand based on image |
| CN103903441A (en) * | 2014-04-04 | 2014-07-02 | 山东省计算中心 | Road traffic state distinguishing method based on semi-supervised learning |
Cited By (27)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106250832A (en) * | 2016-07-25 | 2016-12-21 | 华南理工大学 | A kind of national recognition methods based on integrated convolutional neural networks |
| CN107798281B (en) * | 2016-09-07 | 2021-10-08 | 北京眼神科技有限公司 | Face living body detection method and device based on LBP (local binary pattern) characteristics |
| CN107798281A (en) * | 2016-09-07 | 2018-03-13 | 北京眼神科技有限公司 | A kind of human face in-vivo detection method and device based on LBP features |
| CN106529542A (en) * | 2016-09-30 | 2017-03-22 | 中国石油天然气股份有限公司 | Indicator diagram identification method and device |
| CN106650612A (en) * | 2016-10-27 | 2017-05-10 | 嘉兴学院 | Road vehicle detection and classification method |
| CN108345896A (en) * | 2017-01-23 | 2018-07-31 | 许继集团有限公司 | High-tension apparatus image-recognizing method and device based on variety of components and scale of components |
| CN110494902A (en) * | 2017-02-03 | 2019-11-22 | 西门子交通有限责任公司 | For managing the system, apparatus and method of the traffic in geographical location |
| CN106897695A (en) * | 2017-02-24 | 2017-06-27 | 上海斐讯数据通信技术有限公司 | A kind of image recognizing and processing equipment, system and method |
| CN106971556A (en) * | 2017-05-16 | 2017-07-21 | 中山大学 | The recognition methods again of bayonet vehicle based on dual network structure |
| CN109002830A (en) * | 2017-06-06 | 2018-12-14 | 罗伯特·博世有限公司 | The for motor vehicle method and apparatus for object of classification |
| CN109002830B (en) * | 2017-06-06 | 2024-04-30 | 罗伯特·博世有限公司 | Method and device for classifying objects for vehicles |
| CN108494620A (en) * | 2018-02-28 | 2018-09-04 | 南京邮电大学 | Network service flow feature selecting and sorting technique based on multiple target Adaptive evolvement arithmetic |
| CN108388871A (en) * | 2018-02-28 | 2018-08-10 | 中国计量大学 | A kind of vehicle checking method returned based on vehicle body |
| CN108388871B (en) * | 2018-02-28 | 2021-05-18 | 中国计量大学 | Vehicle detection method based on vehicle body regression |
| CN108494620B (en) * | 2018-02-28 | 2021-07-27 | 南京邮电大学 | Network service flow characteristic selection and classification method |
| CN108810547A (en) * | 2018-07-03 | 2018-11-13 | 电子科技大学 | A kind of efficient VR video-frequency compression methods based on neural network and PCA-KNN |
| CN110858303A (en) * | 2018-08-22 | 2020-03-03 | 西门子（中国）有限公司 | Method and device for detecting license plate number |
| CN109299729A (en) * | 2018-08-24 | 2019-02-01 | 四川大学 | Vehicle checking method and device |
| CN109299729B (en) * | 2018-08-24 | 2021-02-23 | 四川大学 | Vehicle detection method and device |
| CN110135318A (en) * | 2019-05-08 | 2019-08-16 | 佳都新太科技股份有限公司 | Cross determination method, apparatus, equipment and the storage medium of vehicle record |
| CN112036421A (en) * | 2019-05-16 | 2020-12-04 | 搜狗（杭州）智能科技有限公司 | Image processing method and device and electronic equipment |
| CN110598749A (en) * | 2019-08-13 | 2019-12-20 | 广东数鼎科技有限公司 | Image vehicle recognition implementation system based on convolutional neural network AI algorithm |
| US11386680B2 (en) | 2020-03-28 | 2022-07-12 | Wipro Limited | System and method of identifying vehicle brand and model |
| CN113537666A (en) * | 2020-04-16 | 2021-10-22 | 马上消费金融股份有限公司 | Evaluation model training method, evaluation and business auditing method, device and equipment |
| CN113537666B (en) * | 2020-04-16 | 2024-05-03 | 马上消费金融股份有限公司 | Evaluation model training method, evaluation and business auditing method, device and equipment |
| CN112801233A (en) * | 2021-04-07 | 2021-05-14 | 杭州海康威视数字技术股份有限公司 | Internet of things equipment honeypot system attack classification method, device and equipment |
| WO2022247573A1 (en) * | 2021-05-26 | 2022-12-01 | 腾讯科技（深圳）有限公司 | Model training method and apparatus, image processing method and apparatus, device, and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105787437B (en) | 2017-04-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105787437A (en) | Vehicle brand type identification method based on cascading integrated classifier | |
| Huttunen et al. | Car type recognition with deep neural networks | |
| Boukerche et al. | Automated vehicle detection and classification: Models, methods, and techniques | |
| CN102346847B (en) | License plate character recognizing method of support vector machine | |
| KR101395094B1 (en) | Method and system for detecting object in input image | |
| CN105279475A (en) | Fake-licensed vehicle identification method and apparatus based on vehicle identity recognition | |
| CN110781798B (en) | Vehicle-mounted suspect locking system based on raspberry group and face recognition | |
| CN105335757A (en) | Model identification method based on local characteristic aggregation descriptor | |
| CN108830254B (en) | Fine-grained vehicle type detection and identification method based on data balance strategy and intensive attention network | |
| Màrmol et al. | QuickSpot: a video analytics solution for on-street vacant parking spot detection | |
| Chetouane et al. | Vision‐based vehicle detection for road traffic congestion classification | |
| CN110619277A (en) | Multi-community intelligent deployment and control method and system | |
| CN111582178B (en) | Vehicle weight recognition method and system based on multi-azimuth information and multi-branch neural network | |
| CN111078946A (en) | Bayonet vehicle retrieval method and system based on multi-target regional characteristic aggregation | |
| CN105740886A (en) | Machine learning based vehicle logo identification method | |
| CN113177518A (en) | Vehicle weight identification method recommended by weak supervision area | |
| Arth et al. | Object reacquisition and tracking in large-scale smart camera networks | |
| Lu et al. | A new video-based crash detection method: balancing speed and accuracy using a feature fusion deep learning framework | |
| Sikirić et al. | Image representations on a budget: Traffic scene classification in a restricted bandwidth scenario | |
| Outay et al. | Estimating ambient visibility in the presence of fog: a deep convolutional neural network approach | |
| Chen et al. | A visual attention based convolutional neural network for image classification | |
| Peng et al. | Real-time illegal parking detection algorithm in urban environments | |
| He et al. | Vehicle theft recognition from surveillance video based on spatiotemporal attention | |
| Iqbal et al. | Deep learning based smart parking for a metropolitan area | |
| Jiang et al. | Vehicle classification for traffic surveillance videos based on spatial location information and Sparse Representation-based Classifier |
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 |