CN106056086A - Vehicle brand and model identification method based on fast learning framework - Google Patents
Vehicle brand and model identification method based on fast learning framework Download PDFInfo
- Publication number
- CN106056086A CN106056086A CN201610383537.XA CN201610383537A CN106056086A CN 106056086 A CN106056086 A CN 106056086A CN 201610383537 A CN201610383537 A CN 201610383537A CN 106056086 A CN106056086 A CN 106056086A
- Authority
- CN
- China
- Prior art keywords
- feature
- car
- vehicle
- face image
- method based
- 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
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a vehicle brand and model identification method based on a fast learning framework. The vehicle brand and model identification method comprises the steps of: establishing a spark platform, and acquiring a monitoring video stream; detecting a license plate from an obtained video image; acquiring a vehicle face image according to position of the license plate, and carrying out vehicle face recognition; expanding the upper portion, the lower portion, the left portion and the right portion according to width and height of the license plate, and acquiring a vehicle face image; constructing a multi-scale vehicle face image, extracting local binary and gradient direction histogram features of each block among superimposed blocks of the multi-scale vehicle face image, and combining the local binary and gradient direction histogram features into final vehicle face features; training a multi-level cascade classifier, adopts a multi-classifier voting manner at the first level, accepting a result when the majority of classifiers vote to the same target, otherwise, entering a next level; and adopting an integrated classification system at the second level, training a plurality of sub-classifiers through feature mapping, and fusing results of the classifiers to obtain a final classification result. The vehicle brand and model identification method can identify vehicle brands and specific models accurately, quickly and stably in large amount of data.
Description
Technical field
The present invention relates to image steganalysis and intelligent transportation field, more particularly to a kind of based on Fast Learning framework
Vehicle brand type identifier method.
Background technology
Along with the sustainable development of national economy, motor vehicles has become as transportation indispensable in people's daily life
Instrument.But thing followed traffic problems also become increasingly conspicuous.Countries in the world all management to traffic system are stepped up its investment, gradually
Define control of traffic and road research field.
Vehicle cab recognition is the main task in intelligent transportation system and key technology, and it has a wide range of applications, such as highway
Automatic charging, parking lot management, inspection theft vehicle, fake-licensed car management etc..The difficult point of vehicle cab recognition is the general of identification system
All over adaptability, a good vehicle identification system is required to adapt to various scene changes, Changes in weather and vehicle itself
Abrasion etc..
Vehicle identification is concentrated mainly on vehicle cab recognition and vehicle-logo recognition both direction at present.Vehicle cab recognition is to distinguish vehicle kind
Style number is main, and e.g., car, truck, bus etc., its method is the most extensive with identification based on induction coil, but should
Method installs inconvenient, maintenance cost height and discrimination is low.Vehicle-logo recognition can identify vehicle brand, but different brand cars
Mark changes greatly, and car mark is the least so that the detection of car mark is relatively difficult.
Along with the quick growth of transport information stream, its data volume has reached the highest scale.Traditional calculating system face
To such mass data, it is difficult to meet demand.
Chinese patent literature CN 105574543 discloses a kind of vehicle brand type identifier method based on degree of depth study,
Including step 1, training data sets up SVM car plate discrimination model；Step 2, sets up vehicle cab recognition mould based on deep approach of learning training
Type；Step 3, the background of image in video is modeled obtain moving target, pursuit movement Target Acquisition movement objective orbit,
Obtain comprising the picture of car plate；Step 4, is processed described picture by image processing techniques, and acquisition comprises several of car plate
Segment, differentiates with described SVM car plate discrimination model and retains the segment comprising car plate；Step 5, according to the position of car plate, respectively to
Upper and lower, left and right four direction extension setting regions, obtains the band of position of headstock；Step 6, based on described vehicle cab recognition model
Vehicle is identified by the band of position according to described headstock.Although this kind of method utilizes autonomic learning feature, but uses one
Individual grader, classifying quality is poor.And when in the face of mass data, processing speed is slow, and accuracy rate ratio is relatively low.
Summary of the invention
For above-mentioned technical problem, the present invention seeks to: a kind of vehicle brand model based on Fast Learning framework is provided
Recognition methods, can accurately, quickly and stably identify vehicle brand and concrete model in the data of magnanimity.
The technical scheme is that
A kind of vehicle brand type identifier method based on Fast Learning framework, it is characterised in that comprise the following steps:
S01: build spark platform, obtains monitoring video flow；
S02: detect car plate from the video image obtained, according to car plate position acquisition car face image, driving face of going forward side by side is known
Not；
S03: according to width and the height of car plate, extend up and down, it is thus achieved that car face image；
S04: build multiple dimensioned car face image, is had the piecemeal of overlap, and each fritter is extracted partial binary (LBP)
With gradient orientation histogram (HOG) feature, and it is combined into final car face feature；
S05: training multi-stage cascade grader, the first order uses the mode of multi-categorizer ballot, when most graders throw to
During same target, just accept this result, otherwise enter next stage；The second level uses Ensemble classifier system, by Feature Mapping, instruction
Practice multiple sub-classifiers, merge each classifier result, obtain final classification results.
Preferably, described step S01 includes:
S11: receive video flowing by Spark Streaming；
S12: according to time interval, video flowing is divided into discrete RDD data set, this RDD data set is converted into multiple
Sub-RDD data set, every corresponding pictures of sub-RDD data set, every pictures is processed respectively.
Preferably, in described step S03, left and right extends the car plate width of 1.5 times, downwards the car plate height of extension one times,
Extend up the car plate height of 3.5 times.
Preferably, also include after step S03, calculate the angle of inclination of car plate, and rotate image at this angle to car plate
Lower limb is in horizontal direction.
Preferably, also include after step S03, detect car face characteristic point, according to minimum range and location consistency principle
Matching characteristic point, adjusts car face image according to characteristic point position.
Preferably, described adjustment car face image comprises the following steps:
S31: calculate matching characteristic point and the characteristic point institute to be matched straight slope absolute value of shape, if less than 5 degree, then adding
Enter candidate matches point；
S32: calculate the feature description operator of all candidate feature point, fast robust (surf) characteristic point uses four haar
Wavelet character, calculates the Euclidean distance between all feature description operators and the feature description operator of characteristic point to be matched, if
Small distance is less than threshold value d, then this feature point and to be matched some successful match；
S33: calculate the midpoint abscissa x of all match pointsi, i=1,2 ... n, n are coupling number, if μ and σ is respectively it
Average and standard deviation, filter out and all meet μ-3 σ < xi< point of μ+3 σ, and again calculate their median average c, according to c with
The distance of car plate central point adjusts the right boundary of car face.
Preferably, described step S04 comprises the following steps:
S41: car face is normalized to fixed size, is divided into block, and every piece is separated into multiple unit, according to the right unit
To the mode sliding shoe of next unit, calculate the feature of each block；
S42: (x, y) convolution algorithm generate different scale space to use two-dimensional Gaussian kernel G (x, y, σ) and car face image I
Car face image L (x, y, σ), metric space form is expressed as:
L (x, y, σ)=G (x, y, σ) * I (x, y)
Wherein,
S43:LBP feature uses 8 neighborhoods, radius be 1 mode calculate, add up the rectangular histogram of each unit, rectangular histogram bin
Number is 8, all unit LBP features of each piece is connected, forms the feature of each piece.
360 degree of directions are divided into multiple directions block, width by S44: calculate gradient magnitude and the direction of pixel in each unit
Value adds up the gradient orientation histogram of each unit as weight, the feature of all unit of each piece is connected, and is formed every
The feature of individual block, and the final car face feature of this block of formation of connecting with the feature in step S43.
Compared with prior art, the invention have the advantage that
1, the present invention can accurately, quickly and stably identify vehicle brand and concrete model in the data of magnanimity.Identify
Accuracy is high, speed is fast, the suitability is wide.Build multiple dimensioned car face image can express feature at different scale the most efficiently
On, beneficially image recognition；Utilize elasticity distribution formula data set (RDD) in spark that each piecemeal and two features are used and divided
Cloth mode calculates, and which greatly enhances computational efficiency.
2, use multi-stage cascade grader to obtain result, comprehensively utilize the respective advantage of Various Classifiers on Regional, classification can be made
More robust, can obtain more powerful classifying quality.Each grader differentiates that use RDD realizes Distributed Calculation, substantially increases
Computational efficiency.
Accompanying drawing explanation
Below in conjunction with the accompanying drawings and embodiment the invention will be further described:
Fig. 1 is the flow chart of present invention vehicle brand based on Fast Learning framework type identifier method.
Fig. 2 is the vehicle cab recognition flow chart of present invention vehicle brand based on Fast Learning framework type identifier method；
Fig. 3 is that the grader of present invention vehicle brand based on Fast Learning framework type identifier method differentiates flow chart.
Detailed description of the invention
For making the object, technical solutions and advantages of the present invention of greater clarity, below in conjunction with detailed description of the invention and join
According to accompanying drawing, the present invention is described in more detail.It should be understood that these describe the most exemplary, and it is not intended to limit this
Bright scope.Additionally, in the following description, eliminate the description to known features and technology, to avoid unnecessarily obscuring this
The concept of invention.
Embodiment:
A kind of vehicle brand type identifier method based on Fast Learning framework, comprises the following steps:
(1) according to monitoring demand, efficiency of algorithm and machine performance, select to build spark platform, obtain monitoring video flow.
Spark is that a parallel data processes framework, and it can allow big data combine with real time data application, it is possible to use
Spark Streaming processes real time data.
(2) from the video image obtained, car plate is detected.
(3) according to car plate position acquisition car face image, driving face identification of going forward side by side, concretely comprise the following steps:
(3-1) according to width and the height of car plate, extend up and down, it is thus achieved that car face image.
(3-2) calculate the angle of inclination of car plate, and carrousel face is to level.So it is possible to prevent car face image because of monitored
The impact of equipment setting angle and steering direction etc. occurs, and vehicle cab recognition below is produced large effect.
(3-3) detection car face characteristic point, according to minimum range and location consistency principle matching characteristic point.According to characteristic point
Position adjustment car face image.So, car plate detection right boundary location there will be deviation, causes the car face position detected also can
Bigger error occurs, utilizes characteristic point to readjust the border of car face, car face can be made to align.
(3-4) build multiple dimensioned car face image, had the piecemeal of overlap, each fritter is extracted partial binary
And gradient orientation histogram (HOG) feature, and be combined into final car face feature (LBP).In spark, each piecemeal can generate
Respective RDD, it is achieved Distributed Calculation.LBP mainly expresses image texture, and HOG mainly expresses resemblance, and both combine permissible
Well express the feature of car face；The piecemeal having overlap can express local feature, can embody again global property；Multiple dimensioned can
Feature is expressed the most efficiently on different scale, beneficially image recognition；Utilize the elasticity distribution formula data set in spark
(RDD) use distributed way to calculate each piecemeal and two features, which greatly enhances computational efficiency.
(3-5) training two-stage cascade grader, the first order uses the mode of multi-categorizer ballot, when most graders throw to
During same target, just accept this result, otherwise enter next stage；The second level uses Ensemble classifier system, by Feature Mapping, instruction
Practice multiple sub-classifiers, merge each classifier result, obtain final classification results.In spark, it is right that each grader can generate
The RDD answered, it is achieved Distributed Calculation.Do so provides the benefit that: single grader is limited to grader itself, is difficult to reach
To outstanding result, comprehensively utilize the respective advantage of Various Classifiers on Regional, classification more robust can be made；Grader cascades just as spy
As levying cascade, more powerful classifying quality can be obtained；With the feature calculation in (3-4), each grader differentiates that use RDD is real
Existing Distributed Calculation, substantially increases computational efficiency.
As it is shown in figure 1, step (1) specifically includes following steps:
(1) video flowing is received by Spark Streaming.
(2) according to time interval, video flowing is divided into the most discrete RDD data set, owing to there being multi-channel video, often
Individual RDD may comprise a lot of image data, and this RDD is converted into many sub-RDD, every corresponding pictures of sub-RDD.
(3) every pictures is carried out vehicle cab recognition process.
As in figure 2 it is shown, car face identification specifically includes following steps:
(1) a total of 1032 kinds of vehicles in training sample and test sample, this example are prepared.
(2) mode that texture and color combine, positioning licence plate are utilized.
(3) utilize Radon transform, calculate the angle of inclination of car plate, and rotation image is water to car plate lower limb at this angle
Square to.
(4) according to car plate position, left and right extends the car plate width of 1.5 times, and the car plate height of extension one times, upwards expands downwards
Open up the car plate height of 3.5 times, obtain car face image according to this.
(5) calculate rapid robust feature point (surf), find point of symmetry therein, and adjust car face limit according to characteristic point
Boundary.Particularly as follows:
(5-1) amount of calculation matching characteristic point and the characteristic point institute to be matched straight slope absolute value of shape, if less than 5 degree,
Then add candidate matches point.
(5-2) calculating the feature description operator of all candidate feature point, surf characteristic point uses four haar wavelet characters,
It is respectively as follows: horizontal direction value sum, horizontal direction absolute value sum, vertical direction sum, vertical direction absolute value sum.Calculate
Euclidean distance between all feature description operators and the feature description operator of characteristic point to be matched, if minimum range is less than threshold value
D, then this feature point and to be matched some successful match.
(5-3) the midpoint abscissa x of all match points is calculatedi, i=1,2 ... n, n are coupling number, if μ and σ is respectively it
Average and standard deviation.Filter out and all meet μ-3 σ < xi< point of μ+3 σ, and again calculate their median average c, according to c with
The distance of car plate central point adjusts the right boundary of car face.
(6) car face feature extraction, concrete steps include:
(6-1) car face being normalized to 128 × 64 sizes, be divided into 4 × 2 pieces, every piece is separated into 2 × 2 unit.According to
A right unit, to the mode sliding shoe of next unit, calculates the feature of each block.
(6-2) (x, y) convolution algorithm generate different scale space to use two-dimensional Gaussian kernel G (x, y, σ) and car face image I
Car face image L (x, y, σ), metric space form is expressed as:
L (x, y, σ)=G (x, y, σ) * I (x, y)
Wherein, * is convolution algorithm,
This example selects the gaussian kernel of 5 × 5 sizes, σ ∈ { 5,10}.
(6-3) LBP feature uses 8 neighborhoods, radius be 1 mode calculate, add up the rectangular histogram of each unit, rectangular histogram
Bin number is 8.All unit LBP features of each piece are connected, forms the feature of each piece.
(6-4) calculate gradient magnitude and the direction of pixel in each unit, 360 degree of directions are divided into 8 direction blocks, width
Value adds up the gradient orientation histogram of each unit as weight.The feature of all unit of each piece is connected, is formed every
The feature of individual block, and the final feature of this block of formation of connecting with the feature in (6-2).Total characteristic dimension is 7 × 3 × 4 × (8+8)
× 3=4032.
(6-5) in spark, each yardstick car face corresponding a RDD, the sub-RDD of each piece of regeneration, each sub-RDD can
Carry out Distributed Calculation, be greatly improved calculating speed.
(6-6) owing to having the sliding shoe of overlap to extract feature, feature redundancy occurs unavoidably, finally uses principal component analysis
Method (PCA) reduces intrinsic dimensionality.In this example, final dimension is 1200.
As it is shown on figure 3, this example use two-level classifier cascade mode classifying cart face, naturally it is also possible to use three grades or
More.
The first order is voted with three kinds of graders, and three kinds of graders are respectively support vector machine, artificial neural network and random
Forest.When any two grader is identified as same type, just accepts this result, otherwise enter next stage；The second level uses
Rotate forest Ensemble classifier system.By Feature Mapping, generate the feature set of multiple rotation transformation, utilize these feature sets to train
Multiple sub-classifiers, in this example, sub-classifier uses support vector machine, last comprehensive each sub-classifier score, obtains final dividing
Class result.In spark, each grader, by generating corresponding RDD, by Distributed Calculation, quickly obtains result.
It should be appreciated that the above-mentioned detailed description of the invention of the present invention is used only for exemplary illustration or explains the present invention's
Principle, and be not construed as limiting the invention.Therefore, that is done in the case of without departing from the spirit and scope of the present invention is any
Amendment, equivalent, improvement etc., should be included within the scope of the present invention.Additionally, claims purport of the present invention
Within containing the equivalents falling into scope and border or this scope and border whole change and
Modification.
Claims (7)
1. a vehicle brand type identifier method based on Fast Learning framework, it is characterised in that comprise the following steps:
S01: build spark platform, obtains monitoring video flow；
S02: detect car plate from the video image obtained, according to car plate position acquisition car face image, driving face identification of going forward side by side；
S03: according to width and the height of car plate, extend up and down, it is thus achieved that car face image；
S04: build multiple dimensioned car face image, is had the piecemeal of overlap, and each fritter extracts partial binary (LBP) and ladder
Degree direction histogram (HOG) feature, and it is combined into final car face feature；
S05: training multi-stage cascade grader, the first order uses the mode of multi-categorizer ballot, when most graders are thrown to same
During target, just accept this result, otherwise enter next stage；The second level uses Ensemble classifier system, and by Feature Mapping, training is many
Individual sub-classifier, merges each classifier result, obtains final classification results.
Vehicle brand type identifier method based on Fast Learning framework the most according to claim 1, it is characterised in that institute
State step S01 to include:
S11: receive video flowing by Spark Streaming；
S12: according to time interval, video flowing is divided into discrete RDD data set, this RDD data set is converted into many height
RDD data set, every corresponding pictures of sub-RDD data set, every pictures is processed respectively.
Vehicle brand type identifier method based on Fast Learning framework the most according to claim 1, it is characterised in that institute
Stating in step S03, left and right extends the car plate width of 1.5 times, and the car plate height of extension one times, extends up the car of 3.5 times downwards
Board height.
Vehicle brand type identifier method based on Fast Learning framework the most according to claim 1, it is characterised in that step
Also include after rapid S03, calculate the angle of inclination of car plate, and rotate image to car plate lower limb in horizontal direction at this angle.
5., according to the vehicle brand type identifier method based on Fast Learning framework described in claim 1 or 4, its feature exists
In, also include after step S03, detect car face characteristic point, according to minimum range and location consistency principle matching characteristic point, root
Car face image is adjusted according to characteristic point position.
Vehicle brand type identifier method based on Fast Learning framework the most according to claim 5, it is characterised in that institute
State adjustment car face image to comprise the following steps:
S31: calculate matching characteristic point and the characteristic point institute to be matched straight slope absolute value of shape, if less than 5 degree, then adding time
Select match point；
S32: calculate the feature description operator of all candidate feature point, fast robust (surf) characteristic point uses four haar small echos
Feature, calculates the Euclidean distance between all feature description operators and the feature description operator of characteristic point to be matched, if narrow spacing
From less than threshold value d, then this feature point and to be matched some successful match；
S33: calculate the midpoint abscissa x of all match pointsi, i=1,2 ... n, n for coupling number, if μ and σ be respectively its average and
Standard deviation, filters out and all meets μ-3 σ < xi< point of μ+3 σ, and again calculate their median average c, according in c and car plate
The distance of heart point adjusts the right boundary of car face.
Vehicle brand type identifier method based on Fast Learning framework the most according to claim 1, it is characterised in that institute
State step S04 to comprise the following steps:
S41: car face is normalized to fixed size, is divided into block, and every piece is separated into multiple unit, downward according to the right unit
The mode sliding shoe of one unit, calculates the feature of each block；
S42: (x, y) convolution algorithm generate the car face in different scale space to use two-dimensional Gaussian kernel G (x, y, σ) and car face image I
Image L (x, y, σ), metric space form is expressed as:
L (x, y, σ)=G (x, y, σ) * I (x, y)
Wherein,
S43:LBP feature uses 8 neighborhoods, radius be 1 mode calculate, add up the rectangular histogram of each unit, rectangular histogram bin number
It is 8, all unit LBP features of each piece is connected, forms the feature of each piece；
360 degree of directions are divided into multiple directions block by S44: calculate gradient magnitude and the direction of pixel in each unit, and amplitude is made
Add up the gradient orientation histogram of each unit for weight, the feature of all unit of each piece is connected, forms each piece
Feature, and connect with the feature in step S43 formed this block final car face feature.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610383537.XA CN106056086B (en) | 2016-06-02 | 2016-06-02 | Vehicle brand type identifier method based on Fast Learning frame |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610383537.XA CN106056086B (en) | 2016-06-02 | 2016-06-02 | Vehicle brand type identifier method based on Fast Learning frame |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106056086A true CN106056086A (en) | 2016-10-26 |
| CN106056086B CN106056086B (en) | 2019-05-17 |
Family
ID=57173168
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610383537.XA Active CN106056086B (en) | 2016-06-02 | 2016-06-02 | Vehicle brand type identifier method based on Fast Learning frame |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106056086B (en) |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106778560A (en) * | 2016-12-01 | 2017-05-31 | 中科唯实科技（北京）有限公司 | A kind of model recognizing method based on FHOG features and Linear SVM |
| CN106991420A (en) * | 2017-03-27 | 2017-07-28 | 新智认知数据服务有限公司 | A kind of detection method of license plate of the license plate area regression technique based on piecemeal |
| CN107038442A (en) * | 2017-03-27 | 2017-08-11 | 新智认知数据服务有限公司 | A kind of car plate detection and global recognition method based on deep learning |
| CN107085723A (en) * | 2017-03-27 | 2017-08-22 | 新智认知数据服务有限公司 | A kind of characters on license plate global recognition method based on deep learning model |
| CN107221070A (en) * | 2017-05-24 | 2017-09-29 | 广州市银科电子有限公司 | A kind of bill anti-counterfeit discrimination method recognized based on master pattern fluorescent characteristics |
| CN107392119A (en) * | 2017-07-04 | 2017-11-24 | 西北农林科技大学 | A kind of cattle farm monitor video method for real time filtering and system based on Spark Streaming |
| CN108022429A (en) * | 2016-11-04 | 2018-05-11 | 浙江大华技术股份有限公司 | A kind of method and device of vehicle detection |
| CN108154127A (en) * | 2017-12-27 | 2018-06-12 | 天津智芯视界科技有限公司 | A kind of vehicle identification method based on video and radar |
| CN108171971A (en) * | 2017-12-18 | 2018-06-15 | 武汉烽火众智数字技术有限责任公司 | Vehicular real time monitoring method and system based on Spark Streaming |
| CN108268866A (en) * | 2016-12-30 | 2018-07-10 | 纵目科技（上海）股份有限公司 | A kind of vehicle checking method and system |
| CN108319907A (en) * | 2018-01-26 | 2018-07-24 | 腾讯科技（深圳）有限公司 | A kind of vehicle identification method, device and storage medium |
| CN109948610A (en) * | 2019-03-14 | 2019-06-28 | 西南交通大学 | A kind of vehicle fine grit classification method in the video based on deep learning |
| CN110362713A (en) * | 2019-07-12 | 2019-10-22 | 四川长虹电子系统有限公司 | Video monitoring method for early warning and system based on Spark Streaming |
| CN110765861A (en) * | 2019-09-17 | 2020-02-07 | 中控智慧科技股份有限公司 | Unlicensed vehicle type identification method and device and terminal equipment |
| CN110889867A (en) * | 2018-09-10 | 2020-03-17 | 浙江宇视科技有限公司 | Method and device for detecting damaged degree of car face |
| 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 |
| CN113505863A (en) * | 2021-09-08 | 2021-10-15 | 武大吉奥信息技术有限公司 | Image multistage classification method and system based on cascade mean vector comprehensive scoring |
Citations (6)
| 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 |
| CN104820831A (en) * | 2015-05-13 | 2015-08-05 | 沈阳聚德视频技术有限公司 | Front vehicle face identification method based on AdaBoost license plate location |
| US9195908B2 (en) * | 2013-05-22 | 2015-11-24 | Xerox Corporation | Snow classifier context window reduction using class t-scores and mean differences |
| CN105206060A (en) * | 2015-10-14 | 2015-12-30 | 武汉烽火众智数字技术有限责任公司 | Vehicle type recognition device and method based on SIFT characteristics |
| CN105335702A (en) * | 2015-10-15 | 2016-02-17 | 浙江捷尚视觉科技股份有限公司 | Bayonet vehicle type recognition method based on statistical learning |
| US20160148076A1 (en) * | 2014-11-26 | 2016-05-26 | Xerox Corporation | Method and system for automating an image rejection process |
-
2016
- 2016-06-02 CN CN201610383537.XA patent/CN106056086B/en active Active
Patent Citations (6)
| 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 |
| US9195908B2 (en) * | 2013-05-22 | 2015-11-24 | Xerox Corporation | Snow classifier context window reduction using class t-scores and mean differences |
| US20160148076A1 (en) * | 2014-11-26 | 2016-05-26 | Xerox Corporation | Method and system for automating an image rejection process |
| CN104820831A (en) * | 2015-05-13 | 2015-08-05 | 沈阳聚德视频技术有限公司 | Front vehicle face identification method based on AdaBoost license plate location |
| CN105206060A (en) * | 2015-10-14 | 2015-12-30 | 武汉烽火众智数字技术有限责任公司 | Vehicle type recognition device and method based on SIFT characteristics |
| CN105335702A (en) * | 2015-10-15 | 2016-02-17 | 浙江捷尚视觉科技股份有限公司 | Bayonet vehicle type recognition method based on statistical learning |
Non-Patent Citations (1)
| Title |
|---|
| 师永征: "基于分区决策树组合分类器的多目标分类方法", 《电脑知识与技术》 * |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108022429A (en) * | 2016-11-04 | 2018-05-11 | 浙江大华技术股份有限公司 | A kind of method and device of vehicle detection |
| CN108022429B (en) * | 2016-11-04 | 2021-08-27 | 浙江大华技术股份有限公司 | Vehicle detection method and device |
| CN106778560A (en) * | 2016-12-01 | 2017-05-31 | 中科唯实科技（北京）有限公司 | A kind of model recognizing method based on FHOG features and Linear SVM |
| CN108268866A (en) * | 2016-12-30 | 2018-07-10 | 纵目科技（上海）股份有限公司 | A kind of vehicle checking method and system |
| CN108268866B (en) * | 2016-12-30 | 2020-07-10 | 纵目科技（上海）股份有限公司 | Vehicle detection method and system |
| CN106991420A (en) * | 2017-03-27 | 2017-07-28 | 新智认知数据服务有限公司 | A kind of detection method of license plate of the license plate area regression technique based on piecemeal |
| CN107038442A (en) * | 2017-03-27 | 2017-08-11 | 新智认知数据服务有限公司 | A kind of car plate detection and global recognition method based on deep learning |
| CN107085723A (en) * | 2017-03-27 | 2017-08-22 | 新智认知数据服务有限公司 | A kind of characters on license plate global recognition method based on deep learning model |
| CN107221070A (en) * | 2017-05-24 | 2017-09-29 | 广州市银科电子有限公司 | A kind of bill anti-counterfeit discrimination method recognized based on master pattern fluorescent characteristics |
| CN107392119A (en) * | 2017-07-04 | 2017-11-24 | 西北农林科技大学 | A kind of cattle farm monitor video method for real time filtering and system based on Spark Streaming |
| CN108171971A (en) * | 2017-12-18 | 2018-06-15 | 武汉烽火众智数字技术有限责任公司 | Vehicular real time monitoring method and system based on Spark Streaming |
| CN108154127A (en) * | 2017-12-27 | 2018-06-12 | 天津智芯视界科技有限公司 | A kind of vehicle identification method based on video and radar |
| CN108319907A (en) * | 2018-01-26 | 2018-07-24 | 腾讯科技（深圳）有限公司 | A kind of vehicle identification method, device and storage medium |
| CN110889867A (en) * | 2018-09-10 | 2020-03-17 | 浙江宇视科技有限公司 | Method and device for detecting damaged degree of car face |
| CN109948610A (en) * | 2019-03-14 | 2019-06-28 | 西南交通大学 | A kind of vehicle fine grit classification method in the video based on deep learning |
| 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 |
| CN110362713A (en) * | 2019-07-12 | 2019-10-22 | 四川长虹电子系统有限公司 | Video monitoring method for early warning and system based on Spark Streaming |
| CN110362713B (en) * | 2019-07-12 | 2023-06-06 | 四川长虹云数信息技术有限公司 | Video monitoring and early warning method and system based on Spark Streaming |
| CN110765861A (en) * | 2019-09-17 | 2020-02-07 | 中控智慧科技股份有限公司 | Unlicensed vehicle type identification method and device and terminal equipment |
| CN113505863A (en) * | 2021-09-08 | 2021-10-15 | 武大吉奥信息技术有限公司 | Image multistage classification method and system based on cascade mean vector comprehensive scoring |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106056086B (en) | 2019-05-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106056086B (en) | Vehicle brand type identifier method based on Fast Learning frame | |
| Menon et al. | Detection and recognition of multiple license plate from still images | |
| CN100485710C (en) | Method for recognizing vehicle type by digital picture processing technology | |
| CN101859382B (en) | License plate detection and identification method based on maximum stable extremal region | |
| CN104239867B (en) | License plate locating method and system | |
| CN105740886B (en) | A kind of automobile logo identification method based on machine learning | |
| CN103324930B (en) | A kind of registration number character dividing method based on grey level histogram binaryzation | |
| CN105335702B (en) | A kind of bayonet model recognizing method based on statistical learning | |
| CN109101924A (en) | A kind of pavement marking recognition methods based on machine learning | |
| CN104299008A (en) | Vehicle type classification method based on multi-feature fusion | |
| CN104299009A (en) | Plate number character recognition method based on multi-feature fusion | |
| CN103679191A (en) | An automatic fake-licensed vehicle detection method based on static state pictures | |
| Li et al. | Robust vehicle tracking for urban traffic videos at intersections | |
| CN108108703A (en) | Deceleration strip missing detection method, device and electronic equipment | |
| CN105354533A (en) | Bag-of-word model based vehicle type identification method for unlicensed vehicle at gate | |
| CN101369312B (en) | Method and equipment for detecting intersection in image | |
| CN104143091A (en) | Single-sample face recognition method based on improved mLBP | |
| CN111915583A (en) | Vehicle and pedestrian detection method based on vehicle-mounted thermal infrared imager in complex scene | |
| CN104391966A (en) | Typical car logo searching method based on deep learning | |
| Hamdi et al. | Road signs classification by ANN for real-time implementation | |
| Chiu et al. | Real-time traffic light detection on resource-limited mobile platform | |
| CN109255052B (en) | Three-stage vehicle retrieval method based on multiple features | |
| Coronado et al. | Detection and classification of road signs for automatic inventory systems using computer vision | |
| CN104463238B (en) | A kind of automobile logo identification method and system | |
| Ho et al. | Intelligent speed bump system with dynamic license plate recognition |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20201217 Address after: No.88 Zhujiang Road, Xuecheng District, Zaozhuang City, Shandong Province Patentee after: Shandong Weidi Information Technology Co., Ltd Address before: 215000 2-b803 Creative Industry Park, no.328 Xinghu street, Suzhou Industrial Park, Jiangsu Province Patentee before: JIANGSU HUATONG DATA CLOUD TECHNOLOGY Co.,Ltd. |
|
| TR01 | Transfer of patent right |