CN104361343A - Method and device for identifying vehicle types - Google Patents
Method and device for identifying vehicle types Download PDFInfo
- Publication number
- CN104361343A CN104361343A CN201410604350.9A CN201410604350A CN104361343A CN 104361343 A CN104361343 A CN 104361343A CN 201410604350 A CN201410604350 A CN 201410604350A CN 104361343 A CN104361343 A CN 104361343A
- Authority
- CN
- China
- Prior art keywords
- car
- image
- vehicle
- feature
- car face
- 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
- 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
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a method and a device for identifying vehicle types on the basis of video image analysis. The method includes acquiring vehicle images; identifying regions of interest, and extracting vehicle logo and vehicle face features; classifying the vehicle types according to the vehicle logo and vehicle face features. The method and the device have the advantages that a vehicle face identifying technology is combined with a vehicle logo identifying technology, and accordingly the vehicle type identifying accuracy can be improved.
Description
Technical field
The present invention relates to image recognition and traffic safety technology field, in particular to a kind of vehicle type recognition method based on video image analysis and device thereof.
Background technology
Usually vehicle tested the speed or capture, need to adopt the hardware such as velocity radar or ground induction coil, but these methods cannot provide traffic scene information intuitively.Along with ITS (Intelligent Transport System, intelligent transportation system) the constantly bringing forth new ideas and development of technology, supervisory system based on video sensor obtains significant progress, more specifically, vehicle cab recognition based on video image has become a very important research branch, can be obtained the information in traffic scene by video sensor more intuitively.
Vehicle identification can only by identifying the size of vehicle, and be large, medium and small type car by the existing vehicle cab recognition based on video image.Or in some existing technical scheme, vehicle to be classified as Audi, masses, the serial brand vehicle such as beautiful only by vehicle-logo recognition by it, cannot segment the sub-brand name under various brands type, thus the accuracy of vehicle classification and availability lower.Vehicle-logo recognition technology refers to digital picture or video signal flow for research object, by image procossing and automatic identifying method, obtains a kind of practical technique of motor vehicles brand message.Vehicle-logo recognition technology is as the important step of in intelligent transportation system, and its application is more and more subject to people's attention, and its application relates to highway toll, vehicle management, highway are deployed to ensure effective monitoring and control of illegal activities.Such as University Of Qingdao's journal (engineering version) discloses a kind of vehicle-logo location algorithm based on Laws template convolution [the 27th volume the 3rd phase, in September, 2012], it carries out car mark coarse positioning according to car plate and car target priori, again according to the texture information of radiator-grid around car mark, and utilize the filtering of Laws masterplate to remove the interference of radiator-grid around car mark, realize car target location.
Or in some existing other technologies scheme, sorted out vehicle by the identification of car face, but there is the lower problem of accuracy in it equally.Such as China Patent Publication No. be CN102411710A patent document discloses a kind of vehicle type recognition method based on car face feature, it comprises the steps: (1) by CCTV camera collection vehicle image, and carry out pre-service, then detect and be partitioned into the car face image that can characterize type of vehicle, specifically comprising the steps: the vehicle image of (1-1) usage monitoring camera acquisition various; (1-2) Image semantic classification, adopts homomorphic filtering to strengthen picture quality; (1-3) adopt the car face region detection based on car plate positional information and segmentation, be partitioned into the car face region that can characterize vehicle feature; (2) Curvelet wavelet transformation is carried out to vehicle image, to extract the car face eigenmatrix that can characterize vehicle feature; (3) the Curvelet wavelet-based attribute vector of support vector machine classifier to the car face image extracted is adopted to classify, to identify vehicle.
Summary of the invention
In order to accurately identify concrete type of vehicle, the object of the embodiment of the present invention is to provide a kind of vehicle type recognition method based on video image analysis and device thereof.
The embodiment of the present invention realizes by the following technical solutions:
A kind of vehicle type recognition method, comprising:
Collection vehicle image;
Identify area-of-interest, and extract car mark and car face feature;
According to described car mark and car face feature, type of vehicle is sorted out.
Preferably, the step extracting car face feature comprises:
Car face location is carried out in described area-of-interest;
The pre-service of elimination illumination effect is carried out to the image behind car face location;
From through eliminating the textural characteristics and the edge feature that extract car face illumination effect pretreated image respectively, and after the textural characteristics of car face and edge feature are combined, form car face feature.
Preferably, from through eliminating the method extracting the textural characteristics of car face the pretreated image of illumination effect be respectively:
Carrying out size normalization process to through eliminating the pretreated image of illumination effect, extracting the textural characteristics of Gabor wavelet feature as car face.
Preferably, the step of carrying out car face location in described area-of-interest comprises:
In described area-of-interest, carry out License Plate, and coarse positioning is carried out to car face;
To the target of the view data in the yuv format car face image zooming-out Y passage after coarse positioning as image procossing, calculate car face image gradient figure;
Car face image gradient figure is projected to vertical direction, to determine the up-and-down boundary of car face, and car face image gradient figure is projected to horizontal direction, to determine the right boundary of car face.
Preferably, the step extracting car face feature also comprises the treatment step car face feature obtained being carried out to principal component analysis (PCA) PCA (Principal Component Analysis, principal component analysis (PCA)) dimensionality reduction:
Calculate the covariance matrix of car face feature samples, calculate eigenwert and the proper vector of this covariance matrix, and orthonomalization process is carried out to eigenwert and proper vector;
Sort to eigenwert is descending, get front k eigenwert characteristic of correspondence vector and form matrix, then by the eigenwert after sequence and this matrix multiple, obtaining the compression vehicle face feature of k dimension, wherein, k >=1.
Preferably, the step extracting car mark feature comprises:
Vehicle-logo location is carried out in described area-of-interest;
Size normalization and histogram equalization process are carried out to the image after vehicle-logo location, and adopt the edge feature improved to carry out feature extraction to car mark to the image after process, wherein, operation is weighted with the edge feature in the middle of outstanding car mark to the eigenwert in the middle of image, realizes the improvement of edge feature.
Preferably, the step of carrying out vehicle-logo location in described area-of-interest comprises:
In described area-of-interest, coarse positioning is carried out to car mark;
Adopt and carry out horizontal direction filtering based on the bank of filters of texture to the car logo image after coarse positioning, medium filtering is adopted to carry out denoising to image afterwards, binary conversion treatment is being carried out to image and after obtaining binary image, the process of morphology opening operation is carried out to image, and the image after calculation process is projected on vertical direction, the longest continuum after obtaining projection also it can be used as car target up-and-down boundary;
Respectively horizontal filtering and vertical filtering are carried out to the car logo image after coarse positioning, medium filtering is adopted to carry out denoising to image afterwards, after image being carried out to binaryzation and the process of morphology opening operation, connected component labeling is carried out to the image after calculation process, obtain the picture fh (x after horizontal filtering process, y) the picture fv (x and after vertical filtering, y), relatively picture fh (x, y) with picture fv (x, y) quantity of white pixel, and the picture of white pixel negligible amounts is projected to horizontal direction, to determine car target right boundary.
Preferably, the step extracting car mark feature also comprises:
PCA dimension-reduction treatment is carried out to the car mark feature obtained.
Preferably, according to described car mark and car face feature, the step that type of vehicle is sorted out is comprised:
Support vector machines is adopted to carry out training classification, to obtain the first classification results and the second classification results respectively to car face feature and car mark feature respectively;
Described first classification results and the second classification results are mated, if when including the car target type of vehicle that the second classification results sorts out in the first classification results, select this type of vehicle as type of vehicle categorization results; Otherwise, select categorization results the most similar in multiple first classification results as type of vehicle categorization results.
A kind of vehicle type recognition device, comprising:
Collecting unit, for collection vehicle image;
Area-of-interest recognition unit, for identifying area-of-interest;
Feature extraction unit, for extracting car mark and car face feature;
Vehicle type recognition unit, for sorting out type of vehicle according to described car mark and car face feature.
The present invention, by the identification of car face and vehicle-logo recognition two kinds of technology being combined, can improve the accuracy of vehicle type recognition.
Accompanying drawing explanation
The vehicle type recognition method flow schematic diagram that Fig. 1 provides for the embodiment of the present invention;
The car face positioning flow schematic diagram that Fig. 2 provides for the embodiment of the present invention;
The car face feature identification process schematic diagram that Fig. 3 provides for the embodiment of the present invention;
The vehicle-logo location schematic flow sheet that Fig. 4 provides for the embodiment of the present invention.
The realization of the object of the invention, functional characteristics and excellent effect, be described further below in conjunction with specific embodiment and accompanying drawing.
Embodiment
Below in conjunction with the drawings and specific embodiments, technical scheme of the present invention is described in further detail, can better understand the present invention to make those skilled in the art and can be implemented, but illustrated embodiment is not as a limitation of the invention.
As shown in Figure 1, embodiments provide a kind of vehicle type recognition method, comprise the steps:
S10, collection vehicle image;
S20, identify area-of-interest, and extract car mark and car face feature;
S30, according to described car mark and car face feature, type of vehicle to be sorted out.
In the present invention, first by front end camera, vehicle image is gathered, then ROI (the Region Of Interest needing to identify is detected, area-of-interest) region, edge and textural characteristics is adopted to carry out feature extraction to the car mark of vehicle and car face region, finally utilize support vector machine by graphic collection to 200 kinds of vehicle style classifications under 32 brands such as such as Bora 2006 sections, BMW X6, Magotan 2008 sections, Magotan 2012 sections, improve the accuracy of type of vehicle classification with this.
In the present embodiment, the step extracting car face feature comprises:
S201, in described area-of-interest, carry out car face location;
S202, to car face location after image carry out the pre-service of elimination illumination effect;
S203, from through eliminating the textural characteristics and the edge feature that extract car face illumination effect pretreated image respectively, and after the textural characteristics of car face and edge feature are combined, form car face feature.
Preferably, in described step S203, from through eliminating the method extracting the textural characteristics of car face the pretreated image of illumination effect be respectively: carrying out size normalization process to through eliminating the pretreated image of illumination effect, extracting the textural characteristics of Gabor wavelet feature as car face.
Preferably, in described step S201, the step of carrying out car face location in described area-of-interest comprises:
S2011, in described area-of-interest, carry out License Plate, and coarse positioning is carried out to car face;
S2012, to the target of the view data in the yuv format car face image zooming-out Y passage after coarse positioning as image procossing, calculate car face image gradient figure;
S2013, car face image gradient figure to be projected to vertical direction, to determine the up-and-down boundary of car face, and car face image gradient figure is projected to horizontal direction, to determine the right boundary of car face.
In an embodiment of the present invention, as shown in Figure 2, the step performing car face location comprises:
1) License Plate is carried out to the vehicle image collected, then coarse positioning is carried out to car face.
2) in the car face image of coarse positioning, car face is accurately located, because the data collected from video camera are the view data of yuv format, due to only gray level image can be needed, so only extract the target of the view data in Y passage as image procossing herein, calculate car face image gradient figure.
3) gradient map is projected to vertical direction, gradient image is separated in the middle of image, value in each horizontal ordinate and its symmetric coordinates be this projection and its symmetric position projection gray level and minimum value, calculate the average gray of projected image, travel through all projections, find first to be greater than mean value intersection point with last, then the part between these two points is the height component of car face.
4), after the height of car face is determined, the width segments determining car face is needed.Concrete grammar is: gradient map projected to horizontal direction, and determines that the method for car face height is similar, determines the right boundary of car face, the program flow diagram of car face location.
Preferably, in the present embodiment, the step extracting car face feature also comprises the treatment step car face feature obtained being carried out to principal component analysis (PCA) PCA dimensionality reduction:
The covariance matrix of S204, calculating car face feature samples, calculates eigenwert and the proper vector of this covariance matrix, and carries out orthonomalization process to eigenwert and proper vector;
S205, to sort to eigenwert is descending, get front k eigenwert characteristic of correspondence vector and form matrix, then by the eigenwert after sequence and this matrix multiple, obtaining the compression vehicle face feature that k ties up, wherein, k >=1.
In an embodiment of the present invention, as shown in Figure 3, the step performing the identification of car face comprises:
1) Image semantic classification is carried out to the picture f (x, y) behind location, eliminate illumination to the impact of image.Adopt the method for histogram equalization to carry out pre-service to image, expression formula is as follows:
Wherein, the grey scale mapping value of T (k) corresponding to gray level k, n is f (x, y) image pixel summation, the pixel summation of ni to be gray level be i.
2) size normalization is carried out to the image of input, extract Gabor wavelet feature.Gabor characteristic can well the textural characteristics of Description Image, and the expression formula of Two-Dimensional Gabor Wavelets is as follows:
x'＝x cosθ+y sinθ；
y'＝-x sinθ+y cosθ；
Wherein, λ represents small echo length, and θ represents little wave line of propagation,
represent phase offset, σ is standard deviation, and γ is space ratio.In order to improve the computing velocity of convolution, by fast fourier transform (FFT), image is transformed into frequency field from image area, product in frequency field is equivalent in the spatial domain as convolution, after frequency field makes product, need, by invert fast fourier transformation, result to be transformed in spatial domain.Can a series of images be produced by different small echo length and small echo direction, every piece image is divided into some regions, add up mean value and the variance of each area grayscale, composition Gabor characteristic.
3) in order to better describe car face feature, herein except extracting the textural characteristics of car face, also be extracted the edge feature of car face, then texture and edge feature are combined, generate new car face feature, wherein, textural characteristics and edge feature is combined and the known technology grasped for those skilled in the art of the technological means forming new car face feature.
4) in order to reduce calculated amount, need to carry out PCA (Principal Component Analysis, principal component analysis (PCA)) dimensionality reduction to feature.Its concrete steps are as follows: the covariance matrix first calculating sample, calculate eigenwert and the proper vector of this matrix, and orthonomalization process is carried out to eigenwert and proper vector, then sort to eigenwert is descending, get front k eigenwert characteristic of correspondence vector composition matrix, then assemblage characteristic and this matrix multiple, obtains the compressive features of k dimension.
In the present embodiment, preferably, the step extracting car mark feature comprises:
S206, in described area-of-interest, carry out vehicle-logo location;
S207, size normalization and histogram equalization process are carried out to the image after vehicle-logo location, and adopt the edge feature improved to carry out feature extraction to car mark to the image after process, wherein, operation is weighted with the edge feature in the middle of outstanding car mark to the eigenwert in the middle of image, realizes the improvement of edge feature.
Preferably, for described step S206, the step of carrying out vehicle-logo location in described area-of-interest comprises:
S2061, in described area-of-interest, coarse positioning is carried out to car mark;
S2062, the bank of filters based on texture is adopted to carry out horizontal direction filtering to the car logo image after coarse positioning, medium filtering is adopted to carry out denoising to image afterwards, binary conversion treatment is being carried out to image and after obtaining binary image, the process of morphology opening operation is carried out to image, and the image after calculation process is projected on vertical direction, the longest continuum after obtaining projection also it can be used as car target up-and-down boundary;
S2063, respectively horizontal filtering and vertical filtering are carried out to the car logo image after coarse positioning, medium filtering is adopted to carry out denoising to image afterwards, after image being carried out to binaryzation and the process of morphology opening operation, connected component labeling is carried out to the image after calculation process, obtain the picture fh (x after horizontal filtering process, y) the picture fv (x and after vertical filtering, y), relatively picture fh (x, y) with picture fv (x, y) quantity of white pixel, and the picture of white pixel negligible amounts is projected to horizontal direction, to determine car target right boundary.
According to an embodiment of the present invention, as shown in Figure 4, the step performing vehicle-logo location comprises:
1) in order to pick out net result from the similar car face result of top n, vehicle mark is needed to carry out aid identification.First need to position car mark from the car face picture behind location, directly can certainly carry out vehicle-logo location from area-of-interest.In order to reduce sensing range, carry out coarse positioning to car mark, namely the up-and-down boundary of car mark coarse positioning is identical with the up-and-down boundary of car face picture, and right boundary is the right boundary of car plate.
2) in order to obtain car target up-and-down boundary, a kind of bank of filters based on texture is adopted to carry out horizontal direction filtering to the sheet of marking on a map of the car after coarse positioning herein, then medium filtering is adopted to carry out denoising to image, binary conversion treatment is carried out to image, after obtaining binary image, carry out morphology opening operation.Projected on vertical direction by image after process, adding up every a line pixel value is the number of pixels of 255, and continuum the longest after then finding projection, this region is the height component of image.
3) car mark right boundary is determined and to determine that the method for lower boundary put on by car similar, also be likely vertical or reticular texture because car target background texture may be horizontal texture, so need to carry out horizontal filtering and vertical filtering respectively, adopt medium filtering to carry out denoising, binaryzation and the process of morphology opening operation are carried out to image.Connected component labeling is carried out to the image after process, filter out the little connected domain that area is crossed, obtain the picture fh (x after horizontal filtering process, y) the picture fv (x and after vertical filtering, y), fh (x, y) and fv (x is compared, y) white pixel quantity, chooses the few picture of white pixel quantity as selection net result.
4) by the picture after selection, project to horizontal direction, according to the symmetric feature of car mark, after car mark projects to horizontal direction, if there is pixel in the left survey somewhere of car mark axis, so also there will be pixel at its symmetric coordinates place, otherwise can think that this position is not car target boundary position.The concrete practice is as follows: in order to ensure computing velocity, using the position of picture centre position as car mark axis of symmetry, travel through x-axis from left to right until centre position, if find that on the axis left side white pixel point is contained at this place, so find its symmetrical place around whether also containing white pixel point, if containing white pixel point, explanation has symmetry, then right boundary is determined, if do not contained, then continues traversal and finds till axis.。
Preferably, for described step S206, the step extracting car mark feature also comprises:
S2064, PCA dimension-reduction treatment is carried out to the car mark feature obtained.
Particularly, according to an embodiment of the present invention, the concrete steps performing vehicle-logo recognition comprise:
1) picture behind location, needs to carry out size normalization and histogram equalization, eliminates the impact of the extraneous factors such as illumination, adopts the edge feature improved carry out feature extraction to car mark and identified.Make discovery from observation, car target major side feature concentrates on car target center section, so adopt the mode edge feature of weighting to improve, is namely weighted operation to the eigenwert in the middle of image, with the edge feature in the middle of outstanding car mark.
2) PCA dimensionality reduction is carried out to car mark feature.
Preferably, in described step S30, according to described car mark and car face feature, the step that type of vehicle is sorted out is comprised:
S301, employing support vector machines carry out training classification, to obtain the first classification results and the second classification results respectively to car face feature and car mark feature respectively;
S302, described first classification results and the second classification results to be mated, if when including the car target type of vehicle that the second classification results sorts out in the first classification results, select this type of vehicle as type of vehicle categorization results; Otherwise, select categorization results the most similar in multiple first classification results as type of vehicle categorization results.
Particularly, in one embodiment of the present invention, perform car mark and mate with car face to determine that the step of type of vehicle comprises exactly:
1) adopt support vector machine (SVM) to carry out training to the car face feature after compression to classify, can style type belonging to Direct Recognition vehicle after classification, as popular Bora 2006 sections, Toyota's Corolla 2007 sections, Honda CRV2012 money etc.SVM is a kind of machine learning method of Corpus--based Method theory, and the method can well solve small sample training problem.Classification depends primarily on the selection of SVM kernel function K (x, y), and adopt RBF kernel function to classify to sample herein, its expression is:
Wherein x is feature, and y calculates the proper vector generated, and g is threshold value.Due to the very similar situation of car face feature may be there is under different brands, in order to improve the accuracy rate of identification, needing to return the most similar car face of top n after classification and comparing in order to result that is last and vehicle-logo recognition, returning net result.
2) adopt support vector machine (SVM) to carry out support vector machine (SVM) training and classification to the car mark feature of carrying out PCA dimensionality reduction, obtain final classification results.
3) the most similar top n result obtained of the car mark classification results obtained and car face being classified is mated, if containing the result corresponding with vehicle-logo recognition in the middle of the top n result of car face identification: as vehicle-logo recognition result is: popular, and car face recognition result is: Toyota's Corolla 2007 sections, popular Passat 2007 sections, Honda CRV2012 money, then final type of vehicle classification results is: popular Passat 2007 sections.If the result of not mating with car mark, then net result is vehicle the most similar in car face identification top n result.
The embodiment of the present invention also accordingly provides a kind of vehicle type recognition device, comprising:
Collecting unit, for collection vehicle image;
Area-of-interest recognition unit, for identifying area-of-interest;
Feature extraction unit, for extracting car mark and car face feature;
Vehicle type recognition unit, for sorting out type of vehicle according to described car mark and car face feature.
The vehicle type recognition method that this vehicle type recognition device can provide based on above-described embodiment and carry out effectively and accurately Classification and Identification to the concrete vehicle below numerous automobile brand, can specifically with reference to mentioned above for its specific implementation process.
The foregoing is only the preferred embodiments of the present invention; not thereby the scope of the claims of the present invention is limited; every utilize instructions of the present invention and accompanying drawing content to do equivalent structure or equivalent flow process conversion; or be directly or indirectly used in other relevant technical fields, be all in like manner included in scope of patent protection of the present invention.
Claims (10)
1. a vehicle type recognition method, is characterized in that, comprising:
Collection vehicle image;
Identify area-of-interest, and extract car mark and car face feature;
According to described car mark and car face feature, type of vehicle is sorted out.
2. vehicle type recognition method as claimed in claim 1, is characterized in that, the step extracting car face feature comprises:
Car face location is carried out in described area-of-interest;
The pre-service of elimination illumination effect is carried out to the image behind car face location;
From through eliminating the textural characteristics and the edge feature that extract car face illumination effect pretreated image respectively, and after the textural characteristics of car face and edge feature are combined, form car face feature.
3. vehicle type recognition method as claimed in claim 2, is characterized in that, from through eliminating the method extracting the textural characteristics of car face the pretreated image of illumination effect is respectively:
Carrying out size normalization process to through eliminating the pretreated image of illumination effect, extracting the textural characteristics of Gabor wavelet feature as car face.
4. vehicle type recognition method as claimed in claim 2, it is characterized in that, the step of carrying out car face location in described area-of-interest comprises:
In described area-of-interest, carry out License Plate, and coarse positioning is carried out to car face;
To the target of the view data in the yuv format car face image zooming-out Y passage after coarse positioning as image procossing, calculate car face image gradient figure;
Car face image gradient figure is projected to vertical direction, to determine the up-and-down boundary of car face, and car face image gradient figure is projected to horizontal direction, to determine the right boundary of car face.
5. vehicle type recognition method as claimed in claim 2, is characterized in that, also comprise the treatment step car face feature obtained being carried out to principal component analysis (PCA) PCA dimensionality reduction:
Calculate the covariance matrix of car face feature samples, calculate eigenwert and the proper vector of this covariance matrix, and orthonomalization process is carried out to eigenwert and proper vector;
Sort to eigenwert is descending, get front k eigenwert characteristic of correspondence vector and form matrix, then by the eigenwert after sequence and this matrix multiple, obtaining the compression vehicle face feature of k dimension, wherein, k >=1.
6. vehicle type recognition method as claimed in claim 1, is characterized in that, the step extracting car mark feature comprises:
Vehicle-logo location is carried out in described area-of-interest;
Size normalization and histogram equalization process are carried out to the image after vehicle-logo location, and adopt the edge feature improved to carry out feature extraction to car mark to the image after process, wherein, operation is weighted with the edge feature in the middle of outstanding car mark to the eigenwert in the middle of image, realizes the improvement of edge feature.
7. vehicle type recognition method as claimed in claim 6, it is characterized in that, the step of carrying out vehicle-logo location in described area-of-interest comprises:
In described area-of-interest, coarse positioning is carried out to car mark;
Adopt and carry out horizontal direction filtering based on the bank of filters of texture to the car logo image after coarse positioning, medium filtering is adopted to carry out denoising to image afterwards, binary conversion treatment is being carried out to image and after obtaining binary image, the process of morphology opening operation is carried out to image, and the image after calculation process is projected on vertical direction, the longest continuum after obtaining projection also it can be used as car target up-and-down boundary;
Respectively horizontal filtering and vertical filtering are carried out to the car logo image after coarse positioning, medium filtering is adopted to carry out denoising to image afterwards, after image being carried out to binaryzation and the process of morphology opening operation, connected component labeling is carried out to the image after calculation process, obtain the picture fh (x after horizontal filtering process, y) the picture fv (x and after vertical filtering, y), relatively picture fh (x, y) with picture fv (x, y) quantity of white pixel, and the picture of white pixel negligible amounts is projected to horizontal direction, to determine car target right boundary.
8. vehicle type recognition method as claimed in claim 6, is characterized in that, also comprise:
PCA dimension-reduction treatment is carried out to the car mark feature obtained.
9. vehicle type recognition method as claimed in claim 1, is characterized in that, comprise according to described car mark and car face feature to the step that type of vehicle is sorted out:
Support vector machines is adopted to carry out training classification, to obtain the first classification results and the second classification results respectively to car face feature and car mark feature respectively;
Described first classification results and the second classification results are mated, if when including the car target type of vehicle that the second classification results sorts out in the first classification results, select this type of vehicle as type of vehicle categorization results; Otherwise, select categorization results the most similar in multiple first classification results as type of vehicle categorization results.
10. a vehicle type recognition device, is characterized in that, comprising:
Collecting unit, for collection vehicle image;
Area-of-interest recognition unit, for identifying area-of-interest;
Feature extraction unit, for extracting car mark and car face feature;
Vehicle type recognition unit, for sorting out type of vehicle according to described car mark and car face feature.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410604350.9A CN104361343B (en) | 2014-10-30 | 2014-10-30 | Vehicle type recognition method and its device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410604350.9A CN104361343B (en) | 2014-10-30 | 2014-10-30 | Vehicle type recognition method and its device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104361343A true CN104361343A (en) | 2015-02-18 |
| CN104361343B CN104361343B (en) | 2018-08-31 |
Family
ID=52528601
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410604350.9A Active CN104361343B (en) | 2014-10-30 | 2014-10-30 | Vehicle type recognition method and its device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104361343B (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104680149A (en) * | 2015-03-10 | 2015-06-03 | 苏州科达科技股份有限公司 | Method and system for recognizing object type |
| 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 |
| CN105551046A (en) * | 2015-12-17 | 2016-05-04 | 浙江宇视科技有限公司 | Vehicle face location method and device |
| CN106529471A (en) * | 2016-11-08 | 2017-03-22 | 广东安居宝数码科技股份有限公司 | Vehicle face positioning method and system |
| CN106557579A (en) * | 2016-11-28 | 2017-04-05 | 中通服公众信息产业股份有限公司 | A kind of vehicle model searching system and method based on convolutional neural networks |
| CN106650550A (en) * | 2015-10-28 | 2017-05-10 | 中通服公众信息产业股份有限公司 | Vehicle model identification method and vehicle model identification system through image characteristics of vehicle mark and vehicle head |
| CN106650660A (en) * | 2016-12-19 | 2017-05-10 | 深圳市华尊科技股份有限公司 | Vehicle type recognition method and terminal |
| CN107301388A (en) * | 2017-06-16 | 2017-10-27 | 重庆交通大学 | A kind of automatic vehicle identification method and device |
| CN107451511A (en) * | 2016-05-31 | 2017-12-08 | 优信拍（北京）信息科技有限公司 | A kind of system, method based on vehicle information extraction OBD data |
| CN108090429A (en) * | 2017-12-08 | 2018-05-29 | 浙江捷尚视觉科技股份有限公司 | Face bayonet model recognizing method before a kind of classification |
| CN108254748A (en) * | 2018-01-15 | 2018-07-06 | 武汉理工大学 | Inland navigation craft superelevation alarm system and method based on laser ranging and radar image |
| CN108897055A (en) * | 2016-02-24 | 2018-11-27 | 北京君和信达科技有限公司 | A kind of method for controlling radiation source and fast general formula safe examination system |
| 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 |
| 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 |
| CN111062396A (en) * | 2019-11-29 | 2020-04-24 | 深圳云天励飞技术有限公司 | License plate number recognition method and device, electronic equipment and storage medium |
| CN111199228A (en) * | 2019-12-26 | 2020-05-26 | 深圳市芊熠智能硬件有限公司 | License plate positioning method and device |
| CN111222409A (en) * | 2019-11-26 | 2020-06-02 | 北京迈格威科技有限公司 | Vehicle brand labeling method, device and system |
| CN111723601A (en) * | 2019-03-19 | 2020-09-29 | 杭州海康威视数字技术股份有限公司 | Image processing method and device |
| CN112465816A (en) * | 2020-12-17 | 2021-03-09 | 哈尔滨市科佳通用机电股份有限公司 | Method, system and device for detecting damage fault of windshield of railway motor car |
| CN113569927A (en) * | 2021-07-13 | 2021-10-29 | 安徽绿舟科技有限公司 | Man-machine interaction system of unmanned power station |
| CN113963214A (en) * | 2021-10-27 | 2022-01-21 | 广东利通科技投资有限公司 | Vehicle type recognition method, device, system, computer device and storage medium |
| CN115880565A (en) * | 2022-12-06 | 2023-03-31 | 江苏凤火数字科技有限公司 | Neural network-based scraped vehicle identification method and system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101196979A (en) * | 2006-12-22 | 2008-06-11 | 四川川大智胜软件股份有限公司 | Method for recognizing vehicle type by digital picture processing technology |
| CN104112122A (en) * | 2014-07-07 | 2014-10-22 | 叶茂 | Vehicle logo automatic identification method based on traffic video |
-
2014
- 2014-10-30 CN CN201410604350.9A patent/CN104361343B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101196979A (en) * | 2006-12-22 | 2008-06-11 | 四川川大智胜软件股份有限公司 | Method for recognizing vehicle type by digital picture processing technology |
| CN104112122A (en) * | 2014-07-07 | 2014-10-22 | 叶茂 | Vehicle logo automatic identification method based on traffic video |
Non-Patent Citations (4)
| Title |
|---|
| 周宇: "运动车辆车标定位的四阶段法研究", 《重庆交通学院学报》 * |
| 姜谊: "车牌检测及汽车类型分类方法研究", 《中国优秀硕士学位论文全文数据库》 * |
| 姜谊: "车辆检测及汽车类型分类方法研究", 《中国优秀硕士学位论文全文数据库》 * |
| 童雯: "基于PCA和SVM的车标识别方法研究", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104680149B (en) * | 2015-03-10 | 2018-07-03 | 苏州科达科技股份有限公司 | A kind of object type recognition methods and system |
| CN104680149A (en) * | 2015-03-10 | 2015-06-03 | 苏州科达科技股份有限公司 | Method and system for recognizing object type |
| 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 |
| CN106650550A (en) * | 2015-10-28 | 2017-05-10 | 中通服公众信息产业股份有限公司 | Vehicle model identification method and vehicle model identification system through image characteristics of vehicle mark and vehicle head |
| CN105551046A (en) * | 2015-12-17 | 2016-05-04 | 浙江宇视科技有限公司 | Vehicle face location method and device |
| CN105551046B (en) * | 2015-12-17 | 2019-04-30 | 浙江宇视科技有限公司 | Vehicle face-positioning method and device |
| CN108897055A (en) * | 2016-02-24 | 2018-11-27 | 北京君和信达科技有限公司 | A kind of method for controlling radiation source and fast general formula safe examination system |
| CN107451511A (en) * | 2016-05-31 | 2017-12-08 | 优信拍（北京）信息科技有限公司 | A kind of system, method based on vehicle information extraction OBD data |
| CN106529471A (en) * | 2016-11-08 | 2017-03-22 | 广东安居宝数码科技股份有限公司 | Vehicle face positioning method and system |
| CN106529471B (en) * | 2016-11-08 | 2019-08-23 | 广东安居宝数码科技股份有限公司 | Vehicle face-positioning method and system |
| CN106557579B (en) * | 2016-11-28 | 2020-08-25 | 中通服公众信息产业股份有限公司 | Vehicle model retrieval system and method based on convolutional neural network |
| CN106557579A (en) * | 2016-11-28 | 2017-04-05 | 中通服公众信息产业股份有限公司 | A kind of vehicle model searching system and method based on convolutional neural networks |
| CN106650660A (en) * | 2016-12-19 | 2017-05-10 | 深圳市华尊科技股份有限公司 | Vehicle type recognition method and terminal |
| CN107301388A (en) * | 2017-06-16 | 2017-10-27 | 重庆交通大学 | A kind of automatic vehicle identification method and device |
| CN108090429A (en) * | 2017-12-08 | 2018-05-29 | 浙江捷尚视觉科技股份有限公司 | Face bayonet model recognizing method before a kind of classification |
| CN108090429B (en) * | 2017-12-08 | 2020-07-24 | 浙江捷尚视觉科技股份有限公司 | Vehicle type recognition method for graded front face bayonet |
| CN108254748A (en) * | 2018-01-15 | 2018-07-06 | 武汉理工大学 | Inland navigation craft superelevation alarm system and method based on laser ranging and radar image |
| 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 |
| CN111723601A (en) * | 2019-03-19 | 2020-09-29 | 杭州海康威视数字技术股份有限公司 | Image processing 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 |
| CN111222409A (en) * | 2019-11-26 | 2020-06-02 | 北京迈格威科技有限公司 | Vehicle brand labeling method, device and system |
| CN111062396B (en) * | 2019-11-29 | 2022-03-25 | 深圳云天励飞技术有限公司 | License plate number recognition method and device, electronic equipment and storage medium |
| CN111062396A (en) * | 2019-11-29 | 2020-04-24 | 深圳云天励飞技术有限公司 | License plate number recognition method and device, electronic equipment and storage medium |
| CN111199228A (en) * | 2019-12-26 | 2020-05-26 | 深圳市芊熠智能硬件有限公司 | License plate positioning method and device |
| CN111199228B (en) * | 2019-12-26 | 2023-03-28 | 深圳市芊熠智能硬件有限公司 | License plate positioning method and device |
| CN112465816A (en) * | 2020-12-17 | 2021-03-09 | 哈尔滨市科佳通用机电股份有限公司 | Method, system and device for detecting damage fault of windshield of railway motor car |
| CN113569927A (en) * | 2021-07-13 | 2021-10-29 | 安徽绿舟科技有限公司 | Man-machine interaction system of unmanned power station |
| CN113963214A (en) * | 2021-10-27 | 2022-01-21 | 广东利通科技投资有限公司 | Vehicle type recognition method, device, system, computer device and storage medium |
| CN115880565A (en) * | 2022-12-06 | 2023-03-31 | 江苏凤火数字科技有限公司 | Neural network-based scraped vehicle identification method and system |
| CN115880565B (en) * | 2022-12-06 | 2023-09-05 | 江苏凤火数字科技有限公司 | Neural network-based scraped vehicle identification method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104361343B (en) | 2018-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104361343A (en) | Method and device for identifying vehicle types | |
| Huang et al. | Vehicle detection and inter-vehicle distance estimation using single-lens video camera on urban/suburb roads | |
| CN103971097B (en) | Vehicle license plate recognition method and system based on multiscale stroke models | |
| Derpanis et al. | Classification of traffic video based on a spatiotemporal orientation analysis | |
| CN109101924A (en) | A kind of pavement marking recognition methods based on machine learning | |
| CN102411710A (en) | Vehicle type recognition method based on vehicle face features | |
| CN109635733B (en) | Parking lot and vehicle target detection method based on visual saliency and queue correction | |
| CN107704833A (en) | A kind of front vehicles detection and tracking based on machine learning | |
| CN104463220A (en) | License plate detection method and system | |
| CN103544487A (en) | Front car identification method based on monocular vision | |
| CN104766344A (en) | Vehicle detecting method based on moving edge extractor | |
| CN105354547A (en) | Pedestrian detection method in combination of texture and color features | |
| Chen | Road vehicle recognition algorithm in safety assistant driving based on artificial intelligence | |
| CN106778473A (en) | A kind of model recognizing method | |
| CN104063682A (en) | Pedestrian detection method based on edge grading and CENTRIST characteristic | |
| Hsieh et al. | A real-time mobile vehicle license plate detection and recognition for vehicle monitoring and management | |
| Zhang et al. | A front vehicle detection algorithm for intelligent vehicle based on improved gabor filter and SVM | |
| CN102129569A (en) | Equipment and method for detecting object based on multiscale comparison characteristic | |
| Dalka et al. | Vehicle classification based on soft computing algorithms | |
| Kanchev et al. | Blurred image regions detection using wavelet-based histograms and SVM | |
| Sommer et al. | A comprehensive study on object proposals methods for vehicle detection in aerial images | |
| Chen et al. | Context-aware lane marking detection on urban roads | |
| Ktata et al. | License plate localization using Gabor filters and neural networks | |
| Dallalzadeh et al. | Symbolic classification of traffic video shots | |
| Mejía-Iñigo et al. | Color-based texture image segmentation for vehicle detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information |
Address after: 518000, Shenzhen, Guangdong, Nanshan District Nanshan commercial and cultural center Tianli Central Business Plaza (phase two) C-3005 Applicant after: SHENZHEN UNIHZ TECHNOLOGIES CO., LTD. Address before: 518000, Shenzhen, Guangdong, Nanshan District Nanshan commercial and cultural center Tianli Central Business Plaza (phase two) C-3005 Applicant before: UNIHZ Technologies Corp. |
|
| COR | Change of bibliographic data | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |