CN108614985A - A kind of vehicle-logo location method - Google Patents
A kind of vehicle-logo location method Download PDFInfo
- Publication number
- CN108614985A CN108614985A CN201611139492.8A CN201611139492A CN108614985A CN 108614985 A CN108614985 A CN 108614985A CN 201611139492 A CN201611139492 A CN 201611139492A CN 108614985 A CN108614985 A CN 108614985A
- Authority
- CN
- China
- Prior art keywords
- logo
- image
- radiator
- grid
- projection
- 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.)
- Pending
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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a kind of vehicle-logo location methods, the positioning of car plate and the coarse positioning of logo, distinguish radiator-grid type around logo, the interference of radiator-grid is excluded, image block removes background isolated point, and mathematical morphology operates and ask minimum enclosed rectangle, size verification, determine final logo image, this method can not be illuminated by the light that uneven, logo type is different, logo shape differs and the influence of the factors such as surrounding radiator-grid, effectively improve the accuracy rate of vehicle-logo location.
Description
Technical field
The present invention relates to a kind of localization methods, more particularly to a kind of vehicle-logo location method.
Background technology
With economic rapid development, global auto recoverable amount has broken through 1,000,000,000 high pointes, wherein China's automobile guarantee-quantity
Close to 100,000,000.The suddenly rising of automobile quantity, does not only result in discharge pollutants and steeply rises, and also results in urban transportation and seriously surpasses
Load, and then lead to being on the increase for traffic accident quantity.Therefore, traffic jam is alleviated by high-tech means, improves road
Transport capacity, saves human resources, and the intelligent transportation system for reducing traffic accident is come into being.
In intelligent transportation field, current domestic research institution sets foot in and knows in logo mainly based on license plate recognition technology
The research of not aspect is less.And logo is an important attribute of vehicle, is different from car plate, logo is difficult to replace, alter, it
Robber, which robs the fields such as vehicle is traced, vehicles peccancy automatically records, parking lot unattended, bridge crossing automatic charging, to be had widely
Using with great economic value and realistic meaning.
In the prior art, the method for vehicle-logo location mainly has：License plate locating method based on priori is based on template
Matched localization method, based on energy enhancing and shape filtering vehicle-logo location method, based on the vehicle-logo location of principal component analysis
Method etc..Above-mentioned several vehicle-logo location methods have certain feasibility, however many algorithms only consider relatively easy situation, do not have
Consider situations such as background complexity, uneven illumination, size shape differ.Such as：Analysis for car radiation net, due to setting
The radiator-grid of constantly bringing forth new ideas on meter, automobile is different, there is horizontal texture, vertical texture etc., if at a kind of unified algorithm
Reason, then effect is poor.
Invention content
In view of the shortcomings of the prior art, the present invention provides a kind of vehicle-logo location method, and this method can not be illuminated by the light uneven, vehicle
Mark type is different, logo shape differs and the influence of the factors such as surrounding radiator-grid, effectively improves the accuracy rate of vehicle-logo location.
A kind of vehicle-logo location method, includes the following steps：
1) coarse positioning of the positioning of car plate and logo,
2) radiator-grid type around logo is distinguished,
3) interference of radiator-grid is excluded,
4) image block removes background isolated point,
5) mathematical morphology operates and asks boundary rectangle,
6) size verification completes logo fine positioning,
Logo can be positioned through the above steps.
In step 1), the position of car plate is first oriented, using car plate position and the relatively-stationary priori in logo position,
If the car plate navigated to wide high respectively w, h, extraction car plate normal position width is w, is highly the rectangular region image of 3*h
As coarse positioning image.
In step 2), gray processing is carried out to the coarse positioning image of step 1), the image for shooting acquisition contains certain noise,
Smoothing denoising is carried out to image using neighborhood averaging, retains the feature of image, obtains the logo coarse positioning image of smoothing denoising,
Using the Sobel operators for defining horizontal gradient, the logo coarse positioning image of smoothing denoising is filtered, obtains vertical edges
Then edge detection image carries out integral projection, i.e., in vertical direction：
In formula, VgradH(x) it is the value of vertical integral projection, ranging from 0≤x≤Width of x, Width, Height points
Not Wei image width and height, if projection mean value be m, then meet VgradH(n-1) ＜ m and VgradH(n) ＞ m, then it is assumed that occur
Saltus step.Total transition times of calculated level gradient projection, choose appropriate threshold value, you can distinguish radiator-grid be horizontal distribution also
It is vertical distribution.
In step 3), when radiator-grid is horizontal distribution around step 2) detects logo, then ask vertical using Sobel operators
Straight edge detects, and using the image after binaryzation as operation image；
When radiator-grid is vertical distribution around step 2) detects logo, then horizontal edge is asked to examine using Sobel operators
It surveys, in order to exclude the bonnet line of demarcation above logo and the encirclement line up and down of radiator-grid, horizontal direction throwing is carried out after binaryzation
Shadow, the projection value of logo part is always relatively low compared with bonnet line of demarcation and radiator-grid surround the projection value of line after projection, and
After the projection of logo part, there is region of the sub-fraction projection value close to 0 respectively up and down；According to above-mentioned 2 information, water is calculated
The number of non-zero row, is denoted as nrows, total projection value is denoted as sum, then defines threshold value T=sum/nrows, projection value after flat projection
Region decision more than threshold value T is the encirclement line up and down in bonnet line of demarcation and radiator-grid above logo, by this subregion
The value of corresponding pixel is set as 0, obtains operation image.
In step 4), the image for shooting acquisition is easy influencing for the external factor such as the illumination being subject to, shade, after binaryzation
Logo nearby has less amount of background isolated point residual, to eliminate these isolated points, to the operation image obtained in step 3),
Carry out piecemeal, and calculate each piece two-value and, according to adaptive threshold resolution target block and background block, wherein：
M=Width/m (2)
In formula：M is the quantity of piecemeal on width, and Width is the width of image, and m is the threshold value of width
N=Height/n (3)
In formula：N is the quantity of piecemeal on length degree, and Height is the length of image, and n is the threshold value of length；
The size of input picture is Width × Height, is divided into M × N blocks, and every piece is denoted as A (i, j), using it is adaptive come
Find out threshold value mTh：
In formula：mThFor adaptive threshold value, miFor each piecemeal A (i, j) two-value and；
By the condition for judging background：
In formula, A (i, j)=1 indicates that this piecemeal is object block, and A (i, j)=0 indicates that the piecemeal is background block.
If being judged as object block, the white pixel point in reserved block；If being judged as background block, the white pixel in block
Point is background isolated point, as operation image after rejecting.
In step 5), select rectangular configuration element, closed operation done to the operation image obtained in step 4), obtain one or
It is multiple independent, there is not the connected region of handing-over with the external world, to each connected region, seeks the external square for surrounding its minimum area
Shape.
In step 6), size verification is carried out to the boundary rectangle in step 5)：If minimum enclosed rectangle is unsatisfactory for any bar
Part is then given up；
If boundary rectangle meets the ratio of width to height threshold value and area size threshold value of setting, judge that the boundary of the boundary rectangle is
Logo fine positioning is completed on the boundary of logo.
Advantageous effect
The present invention provides a kind of vehicle-logo location method, this method can not be illuminated by the light uneven, logo type difference, logo shape
The influence that shape differs with factors such as surrounding radiator-grids effectively improves the accuracy rate of vehicle-logo location.
Description of the drawings
Fig. 1 is the flow chart of vehicle-logo location.
Specific implementation mode
With reference to embodiment and attached drawing, the present invention is described in further detail.
Embodiment
A kind of vehicle-logo location method, includes the following steps：
1) coarse positioning of the positioning of car plate and logo,
2) radiator-grid type around logo is distinguished,
3) interference of radiator-grid is excluded,
4) image block removes background isolated point,
5) mathematical morphology operates and asks boundary rectangle,
6) size verification completes logo fine positioning,
Logo can be positioned through the above steps.
In step 1), the position of car plate is first oriented, using car plate position and the relatively-stationary priori in logo position,
If the car plate navigated to wide high respectively w, h, extraction car plate normal position width is w, is highly the rectangular region image of 3*h
As coarse positioning image.
In step 2), gray processing is carried out to the coarse positioning image of step 1), the image for shooting acquisition contains certain noise,
Smoothing denoising is carried out to image using neighborhood averaging, retains the feature of image, obtains the logo coarse positioning image of smoothing denoising,
Using the Sobel operators for defining horizontal gradient, the logo coarse positioning image of smoothing denoising is filtered, obtains vertical edges
Then edge detection image carries out integral projection, i.e., in vertical direction：
In formula, VgradH(x) it is the value of vertical integral projection, ranging from 0≤x≤Width of x, Width, Height points
Not Wei image width and height, if projection mean value be m, then meet VgradH(n-1) ＜ m and VgradH(n) ＞ m, then it is assumed that occur
Saltus step.Total transition times of calculated level gradient projection, choose appropriate threshold value, you can distinguish radiator-grid be horizontal distribution also
It is vertical distribution.
In step 3), when radiator-grid is horizontal distribution around step 2) detects logo, then ask vertical using Sobel operators
Straight edge detects, and using the image after binaryzation as operation image；
When radiator-grid is vertical distribution around step 2) detects logo, then horizontal edge is asked to examine using Sobel operators
It surveys, in order to exclude the bonnet line of demarcation above logo and the encirclement line up and down of radiator-grid, horizontal direction throwing is carried out after binaryzation
Shadow, the projection value of logo part is always relatively low compared with bonnet line of demarcation and radiator-grid surround the projection value of line after projection, and
After the projection of logo part, there is region of the sub-fraction projection value close to 0 respectively up and down；According to above-mentioned 2 information, water is calculated
The number of non-zero row, is denoted as nrows, total projection value is denoted as sum, then defines threshold value T=sum/nrows, projection value after flat projection
Region decision more than threshold value T is the encirclement line up and down in bonnet line of demarcation and radiator-grid above logo, by this subregion
The value of corresponding pixel is set as 0, obtains operation image.
In step 4), the image for shooting acquisition is easy influencing for the external factor such as the illumination being subject to, shade, after binaryzation
Logo nearby has less amount of background isolated point residual, to eliminate these isolated points, to the operation image obtained in step 3),
Carry out piecemeal, and calculate each piece two-value and, according to adaptive threshold resolution target block and background block, wherein：
M=Width/m (2)
In formula：M is the quantity of piecemeal on width, and Width is the width of image, and m is the threshold value of width
N=Height/n (3)
In formula：N is the quantity of piecemeal on length degree, and Height is the length of image, and n is the threshold value of length；
The size of input picture is Width × Height, is divided into M × N blocks, and every piece is denoted as A (i, j), using it is adaptive come
Find out threshold value mTh：
In formula：mThFor adaptive threshold value, miFor each piecemeal A (i, j) two-value and；
By the condition for judging background：
In formula, A (i, j)=1 indicates that this piecemeal is object block, and A (i, j)=0 indicates that the piecemeal is background block,
If being judged as object block, the white pixel point in reserved block；If being judged as background block, the white pixel in block
Point is background isolated point, as operation image after rejecting.
In step 5), select rectangular configuration element, closed operation done to the operation image obtained in step 4), obtain one or
It is multiple independent, there is not the connected region of handing-over with the external world, to each connected region, seeks the external square for surrounding its minimum area
Shape.
In step 6), size verification is carried out to the boundary rectangle in step 5)：If minimum enclosed rectangle is unsatisfactory for any bar
Part is then given up；
If boundary rectangle meets the ratio of width to height threshold value and area size threshold value of setting, judge that the boundary of the boundary rectangle is
Logo fine positioning is completed on the boundary of logo.
As shown in Figure 1：The implementation steps of vehicle-logo location are as follows：
S101 starts：It is ready for vehicle-logo location work；
S102 image captures；
The positioning of S103 car plates and the coarse positioning of logo；
S104 distinguishes radiator-grid type around logo：If radiator-grid is horizontal distribution, S105 is thened follow the steps；If radiator-grid
For vertical distribution, S106 is thened follow the steps；
S105 elimination of level is distributed radiator-grid, redirects and executes step S107；
S106 eliminates vertical distribution radiator-grid, redirects and executes step S107；
S107 removes background isolated point；
S108 logos are accurately positioned；
S109 exports positioning result；
S110 positioning terminates.
Claims (4)
1. a kind of vehicle-logo location method, which is characterized in that include the following steps：
1) coarse positioning of the positioning of car plate and logo,
2) radiator-grid type around logo is distinguished,
3) interference of radiator-grid is excluded,
4) image block removes background isolated point,
5) mathematical morphology operates and asks boundary rectangle,
6) size verification completes logo fine positioning,
Logo can be positioned through the above steps.
2. vehicle-logo location method according to claim 1, which is characterized in that in step 2), to the coarse positioning figure of step 1)
As carrying out gray processing, the image for shooting acquisition contains certain noise, carries out smoothing denoising to image using neighborhood averaging, protects
The feature for staying image obtains the logo coarse positioning image of smoothing denoising, using the Sobel operators for defining horizontal gradient, to smooth
The logo coarse positioning image of denoising is filtered, and is obtained vertical edge detection image, is then accumulated in vertical direction
Divide projection, i.e.,：
Total transition times of calculated level gradient projection choose appropriate threshold value, you can distinguishing radiator-grid is horizontal distribution or hangs down
Straight distribution.
3. vehicle-logo location method according to claim 1, which is characterized in that in step 3), when step 2) detects logo
When surrounding radiator-grid is horizontal distribution, then vertical edge is asked to detect using Sobel operators, and using the image after binaryzation as behaviour
Make image；
When radiator-grid is vertical distribution around step 2) detects logo, then asks horizontal edge to detect using Sobel operators, be
The encirclement line up and down in the bonnet line of demarcation and radiator-grid above logo is excluded, horizontal direction projection is carried out after binaryzation, throws
The projection value of movie queen's logo part is always relatively low compared with bonnet line of demarcation and radiator-grid surround the projection value of line, and logo portion
After point projection, there is region of the sub-fraction projection value close to 0 respectively up and down；According to above-mentioned 2 information, calculated level projection
The number of non-zero row afterwards is denoted as nrows, and total projection value is denoted as sum, then defines threshold value T=sum/nrows, and projection value is more than threshold
The region decision of value T is the encirclement line up and down in the bonnet line of demarcation and radiator-grid above logo, this subregion is corresponding
The value of pixel is set as 0, obtains operation image.
4. vehicle-logo location method according to claim 1, which is characterized in that in step 4), the image for shooting acquisition is easy
By the external factor such as illumination, shade influenced, the logo after binaryzation nearby has less amount of background isolated point residual,
To eliminate these isolated points, to the operation image obtained in step 3), carry out piecemeal, and calculate each piece two-value and, according to
Adaptive threshold resolution target block and background block, wherein：
M=Width/m (2)
The size of N=Height/n (3) input picture is Width × Height, is divided into M × N blocks, and every piece is denoted as A (i, j),
Threshold value m is found out using adaptiveTh：
By the condition for judging background：
If being judged as object block, the white pixel point in reserved block；If being judged as background block, the white pixel point in block is
Background isolated point, as operation image after rejecting.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611139492.8A CN108614985A (en) | 2016-12-12 | 2016-12-12 | A kind of vehicle-logo location method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611139492.8A CN108614985A (en) | 2016-12-12 | 2016-12-12 | A kind of vehicle-logo location method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108614985A true CN108614985A (en) | 2018-10-02 |
Family
ID=63643619
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611139492.8A Pending CN108614985A (en) | 2016-12-12 | 2016-12-12 | A kind of vehicle-logo location method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108614985A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111860474A (en) * | 2019-04-18 | 2020-10-30 | 四川大学 | Car logo positioning method based on background cooling fin ablation |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103077384A (en) * | 2013-01-10 | 2013-05-01 | 北京万集科技股份有限公司 | Method and system for positioning and recognizing vehicle logo |
| CN103093229A (en) * | 2013-01-21 | 2013-05-08 | 信帧电子技术（北京）有限公司 | Positioning method and device of vehicle logo |
| CN204347862U (en) * | 2014-12-01 | 2015-05-20 | 王云 | A kind of vehicle vehicle-logo recognition system |
| CN105023002A (en) * | 2015-07-20 | 2015-11-04 | 上海应用技术学院 | Vehicle logo positioning method based on active vision |
| CN105426896A (en) * | 2015-11-16 | 2016-03-23 | 成都神州数码索贝科技有限公司 | Car logo automatic identification method and system |
| CN105701487A (en) * | 2014-11-26 | 2016-06-22 | 王云 | Vehicle logo positioning method |
-
2016
- 2016-12-12 CN CN201611139492.8A patent/CN108614985A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103077384A (en) * | 2013-01-10 | 2013-05-01 | 北京万集科技股份有限公司 | Method and system for positioning and recognizing vehicle logo |
| CN103093229A (en) * | 2013-01-21 | 2013-05-08 | 信帧电子技术（北京）有限公司 | Positioning method and device of vehicle logo |
| CN105701487A (en) * | 2014-11-26 | 2016-06-22 | 王云 | Vehicle logo positioning method |
| CN204347862U (en) * | 2014-12-01 | 2015-05-20 | 王云 | A kind of vehicle vehicle-logo recognition system |
| CN105023002A (en) * | 2015-07-20 | 2015-11-04 | 上海应用技术学院 | Vehicle logo positioning method based on active vision |
| CN105426896A (en) * | 2015-11-16 | 2016-03-23 | 成都神州数码索贝科技有限公司 | Car logo automatic identification method and system |
Non-Patent Citations (3)
| Title |
|---|
| 孙翼: "车标自识别系统算法研究", 《中国优秀硕士学位论文全文数据库 工程科技II辑》 * |
| 曹蒙蒙: "基于图像内容的交通车辆识别关键技术", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 魏平顺: "智能交通系统中车标图像识别技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111860474A (en) * | 2019-04-18 | 2020-10-30 | 四川大学 | Car logo positioning method based on background cooling fin ablation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103077384B (en) | A kind of method and system of vehicle-logo location identification | |
| CN104036262B (en) | A kind of method and system of LPR car plates screening identification | |
| CN104732227B (en) | A kind of Location Method of Vehicle License Plate based on definition and luminance evaluation | |
| CN105260713B (en) | A kind of method for detecting lane lines and device | |
| CN104933409B (en) | A kind of parking stall recognition methods based on panoramic picture dotted line feature | |
| CN102419820A (en) | Method for rapidly detecting car logo in videos and images | |
| CN104778444B (en) | The appearance features analysis method of vehicle image under road scene | |
| CN105913685A (en) | Video surveillance-based carport recognition and intelligent guide method | |
| CN102999753B (en) | License plate locating method | |
| CN109784344A (en) | A kind of non-targeted filtering method of image for ground level mark identification | |
| CN102999749B (en) | Based on the securing band violation event intelligent detecting method of Face datection | |
| Li et al. | Nighttime lane markings recognition based on Canny detection and Hough transform | |
| CN110414355A (en) | The right bit sky parking stall of view-based access control model and parking stall line detecting method during parking | |
| CN103268471A (en) | Vehicle illegal land occupying detection method and device | |
| Guo et al. | Image-based seat belt detection | |
| CN108256521A (en) | For the effective coverage localization method of body color identification | |
| CN108009522A (en) | A kind of Approach for road detection, device and terminal | |
| CN105740836B (en) | A kind of illegal detection method for occupying Emergency Vehicle Lane | |
| CN107895151A (en) | Method for detecting lane lines based on machine vision under a kind of high light conditions | |
| CN102938057B (en) | A kind of method for eliminating vehicle shadow and device | |
| CN105741559A (en) | Emergency vehicle lane illegal occupation detection method based on lane line model | |
| CN103927548B (en) | Novel vehicle collision avoiding brake behavior detection method | |
| CN105117726A (en) | License plate positioning method based on multi-feature area accumulation | |
| CN103544480A (en) | Vehicle color recognition method | |
| CN105389561B (en) | A kind of bus zone detection method based on video |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181002 |