CN110032991A - A kind of logo detection and recognition methods based on logo repositioning - Google Patents
A kind of logo detection and recognition methods based on logo repositioning Download PDFInfo
- Publication number
- CN110032991A CN110032991A CN201910331226.2A CN201910331226A CN110032991A CN 110032991 A CN110032991 A CN 110032991A CN 201910331226 A CN201910331226 A CN 201910331226A CN 110032991 A CN110032991 A CN 110032991A
- Authority
- CN
- China
- Prior art keywords
- logo
- image
- repositioning
- license plate
- pixel
- 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
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
- G06V20/625—License plates
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to a kind of logo detections and recognition methods based on logo repositioning.Color characteristic is used first to position license plate as the location feature of license plate, logo can substantially be positioned by the license plate location information oriented.After frame selects the coarse positioning range of logo, the present invention further positions logo by a kind of new logo extracting method based on Canny operator.After the completion of positioning, since vehicle radiator-grid is different, some situations can not be accurately positioned, the present invention proposes a kind of thought based on logo repositioning, relocates logo classification range.Since logo edge feature is obvious, the present invention uses gradient orientation histogram (HOG) feature as characteristic of division.Gradient orientation histogram (HOG) feature of the logo range extracted is calculated, support vector machines (SVM) is sent into and is trained classification.The present invention can effectively detect logo image.
Description
Technical field
The present invention relates to pattern-recognition and computer vision field, especially a kind of logo detection based on logo repositioning
And recognition methods.
Background technique
Intelligent transportation, core concept are that with sensor, the modes such as camera acquire car data, and use computer aided manufacturing
It helps the mode of management to replace traditional direct surveillance, and data sharing and retrieval can be rapidly completed, reach traffic administration integration
Effect.Among these, computer vision technique plays vital task.Complete times that intelligent transportation is endowed
Business, basis is that the acquisition of car data.And for an automobile, license plate and logo are two important marks
Will.To Car license recognition, there are various researchs both at home and abroad, also there are a large amount of mature and stable systems to be developed, and
It is widely used in actual production scene.But the research of vehicle-logo recognition is but paid little attention to, and this is in intelligent transportation
A ring.Therefore, the demand of vehicle-logo recognition is increasingly being increased in the future.
Summary of the invention
In view of this, the purpose of the present invention is to propose to a kind of logo detection and recognition methods based on logo repositioning, energy
It is enough that effectively logo is positioned and detected.
The present invention is realized using following scheme: a kind of logo detection and recognition methods based on logo repositioning, including with
Lower step:
Step S1: the direct picture of variety classes vehicle is obtained, and the direct picture is renamed to mark
The direct picture is divided into two parts by the classification of vehicle in proportion, wherein 2/3 is used as training set, residue 1/3 is used as test set;
Step S2: license plate area present in the test set direct picture is extracted using hsv color space law, and is utilized
Morphological transformation algorithm detection screening license plate area;
Step S3: the license plate area filtered out by step S2 determines logo range, is then obtained using the method for repositioning
Take the car mark region of repositioning;
Step S4: Training Support Vector Machines extract histograms of oriented gradients to the car mark region of repositioning, and use is trained
Support vector machines is classified.
Further, the step S2 specifically includes the following steps:
Step S21: the color space of the direct picture is passed through in openCV image vision library by RGB color
CvtColor function is converted to hsv color space;
Step S22: search step S21 is converted to each of the image in hsv color space pixel, if currently searching
Three channels of rope pixel meet respectively: the i.e. H range in form and aspect channel is between 90-150, and purity channel, that is, S range is in 55-
Between 255, current pixel point value is then 1, is otherwise just 0, finally obtains one by lightness channel, that is, V between 30-255
Pixel value only has 0 and 1 black white image;
Step S23: opening operation is carried out to the black white image that step S22 is obtained using the operator that size is [4,4], can be gone
Except noise;
Step S24: carrying out closed operation to the step S23 image for having removed noise using the operator that size is [20,2], can
To connect the region not being filled, a possibility that there are license plates region is obtained;
Step S25: carrying out opening operation to the license plate possibility image that step S24 is obtained using the operator that size is [1,5],
The edge that the license plate possibility region that S24 is obtained can be removed is short-tempered;
Step S26: the black white image obtained by step S23, step S24, the complete S22 of step S25 step process, final
To each connected region, i.e. license plate possibility region, the ratio of width to height and area of each connected region are calculated;According to formula:
Wherein WHi is the length-width ratio of i-th of connected region, and Area is the area of the connected region;Meet above-mentioned formula then
It is determined as license plate area.
Further, the step S3 specifically includes the following steps:
Step S31: coarse positioning is carried out to logo, positioning relation is obtained with following formula:
Wherein, lb, rb, tb, bb respectively correspond the left and right up-and-down boundary of the thick range of logo;Lp, rp, bp, tp are respectively corresponded
The left and right up-and-down boundary of license plate;
Step S32: the thick range of logo is handled using Canny edge detection algorithm, obtains the thick volume edges of logo
Image;
Step S33: the thick volume edges image of logo is obtained to S32 using the operator that size is [1,2] and carries out opening operation, is used
So that the pixel at logo edge remains;
Step S34: in the logo edge remained due to step S33, some pixels not belong to logo, use system
Meter other pixel point methods of neighborhood of pixels determine whether the pixel that step S33 is remained is logo pixel;I.e. statistics is every
Whether the reservation number of pixels in the neighborhood of a pixel 7*7 is greater than 30;If number is considered as noise pixel less than 30,
Given up;Otherwise retain;
Step S35: the step S34 pixel finally remained is put into the original position of pixel, is finally protected
The image left carries out closed operation using the operator that size is [15,1] to the image finally retained, merges step S34 and retain
The pixel to get off;
Step S36: repositioning car mark region, and formula is as follows:
Wherein, nlb, nrb, ntb, nbb respectively correspond the left and right up-and-down boundary of logo repositioning range, lp, rp, bp, tp
Respectively correspond the left and right up-and-down boundary of license plate.
Further, the step S4 specifically includes the following steps:
Step S41: size normalization is carried out to the logo range of repositioning, is uniformly scaled 64*64;
Step S42: openCV image vision library is used to normalized repositioning logo range image
HOGDescriptor class extracts histograms of oriented gradients;Wherein block size is 16*16, and cell element size is 8*8, and block slides increment
For 8*8, gradient direction number is 9, and the characteristic dimension finally extracted is 1764 dimensions;
Step S43: the training sample image in the training set described in step S1 carries out the detection and extraction of logo, will be correct
The logo of extraction filters out, and is labeled as generic, concurrently sets the negative class of a non-logo type;Vector machine type is C
Class support vector machines set kernel function as POLY function;With the logo training classifier marked;
Step S44: classification is re-started with trained classifier training set described in step S1, by erroneous detection positive sample
Non- logo image tagged be negative class, be fed back to classifier and re-start training, to enable classifying quality to be promoted.
Compared with prior art, the invention has the following beneficial effects:
1, the present invention can effectively detect logo image.
2, the logo background suppression method energy effective filter background information proposed by the present invention based on Canny operator, mentions
It picks up the car and marks main body.
3, the method that the present invention proposes logo repositioning chooses logo range to be sorted, and vehicle is not only utilized in this method
The information of sample body also utilizes the background information of vehicle radiator-grid.
4, the characteristics of concentrating for logo edge energy, the present invention are extracted special using gradient orientation histogram (HOG) algorithm
Sign.HOG is the algorithm based on edge feature, and logo often has obviously edge feature, therefore gradient direction is straight
Side figure (HOG) algorithm to vehicle-logo recognition task have robustness, be very suitable for using.Using support vector machines (SVM) classifier
It arranges in pairs or groups therewith, the effect reached can be more than all similar algorithms.
Detailed description of the invention
Fig. 1 is the flow chart of the embodiment of the present invention.
Specific embodiment
The present invention will be further described with reference to the accompanying drawings and embodiments.
As shown in Figure 1, a kind of logo detection and recognition methods based on logo repositioning is present embodiments provided, it is specific to wrap
Include following steps:
Step S1: the direct picture of variety classes vehicle is obtained, and the direct picture is renamed to mark
The direct picture is divided into two parts by the classification of vehicle in proportion, wherein 2/3 is used as training set, residue 1/3 is used as test set;
Step S2: license plate area present in the test set direct picture is extracted using hsv color space law, and is utilized
Morphological transformation algorithm detection screening license plate area;
Step S3: the license plate area filtered out by step S2 determines logo range, is then obtained using the method for repositioning
Take the car mark region of repositioning;
Step S4: Training Support Vector Machines extract histograms of oriented gradients to the car mark region of repositioning, and use is trained
Support vector machines is classified.
In the present embodiment, the step S1 specifically includes the following contents:
Website crawlers are constructed, a large amount of vehicle datas is obtained, is marked with image name different classes of;I.e. from website
The direct picture for crawling 7 class vehicles, image is renamed, wherein the i-th class image file name is started with digital i, in order to subsequent
Classification;
In the present embodiment, step S2 specifically includes the following steps:
Step S21: the color space of the direct picture comprising logo is passed through into openCV image vision by RGB color
CvtColor function in library is converted to hsv color space;
Step S22: search step S21 is converted to each of the image in hsv color space pixel, if currently searching
Three channels of rope pixel meet respectively: the i.e. H range in form and aspect channel is between 90-150, and purity channel, that is, S range is in 55-
Between 255, current pixel point value is then 1, is otherwise just 0, finally obtains one by lightness channel, that is, V between 30-255
Pixel value only has 0 and 1 black white image;
Step S23: opening operation is carried out to the black white image that step S22 is obtained using the operator that size is [4,4], can be gone
Except noise;
Step S24: carrying out closed operation to the step S23 image for having removed noise using the operator that size is [20,2], can
To connect the region not being filled, a possibility that there are license plates region is obtained;
Step S25: carrying out opening operation to the license plate possibility image that step S24 is obtained using the operator that size is [1,5],
The edge that the license plate possibility region that S24 is obtained can be removed is short-tempered;
Step S26: the black white image obtained by step S23, step S24, the complete S22 of step S25 step process, final
To each connected region, i.e. license plate possibility region, the ratio of width to height and area of each connected region are calculated;According to formula:
Wherein WHi is the length-width ratio of i-th of connected region, and Area is the area of the connected region;Meet above-mentioned formula then
It is determined as license plate area.
In the present embodiment, step S3 specifically includes the following steps:
Step S31: coarse positioning is carried out to logo, positioning relation is obtained with following formula:
Wherein, lb, rb, tb, bb respectively correspond the left and right up-and-down boundary of the thick range of logo;Lp, rp, bp, tp are respectively corresponded
The left and right up-and-down boundary of license plate;
Step S32: the thick range of logo is handled using Canny edge detection algorithm, obtains the thick volume edges of logo
Image;
Step S33: the thick volume edges image of logo is obtained to S32 using the operator that size is [1,2] and carries out opening operation, is used
So that the pixel at logo edge remains;
Step S34: in the logo edge remained due to step S33, some pixels not belong to logo, the present invention
Using statistical pixel neighborhood, other pixel point methods determine whether the pixel that step S33 is remained is logo pixel;I.e.
Whether the reservation number of pixels counted in the neighborhood of each pixel 7*7 is greater than 30;If number is considered as noise less than 30
Pixel is given up;Otherwise retain;
Step S35: the step S34 pixel finally remained is put into the original position of pixel, is finally protected
The image left carries out closed operation using the operator that size is [15,1] to the image finally retained, can merge step S34
The pixel remained;
Step S36: repositioning car mark region, and formula is as follows:
Wherein, nlb, nrb, ntb, nbb respectively correspond the left and right up-and-down boundary of logo repositioning range, lp, rp, bp, tp
Respectively correspond the left and right up-and-down boundary of license plate.
In the present embodiment, step S4 specifically includes the following steps:
Step S41: size normalization is carried out to the logo range of repositioning, is uniformly scaled 64*64;
Step S42: openCV image vision library is used to normalized repositioning logo range image
HOGDescriptor class extracts histograms of oriented gradients;Wherein block size is 16*16, and cell element size is 8*8, and block slides increment
For 8*8, gradient direction number is 9, and the characteristic dimension finally extracted is 1764 dimensions;
Step S43: the training sample image in the training set described in step S1 carries out the detection and extraction of logo, will be correct
The logo of extraction filters out, and is labeled as generic, concurrently sets the negative class of a non-logo type;Used vector machine class
Type is C class support vector machines, sets kernel function as POLY function;With the logo training classifier marked;
Step S44: re-starting classification with trained support vector machine classifier training set described in step S1, will
The non-logo image tagged of erroneous detection positive sample is negative class, is fed back to classifier and re-starts training, to make classifying quality
It is promoted.
The present embodiment effectively can be positioned and be detected to logo.Hsv color spatial filtering license plate has been used first
Region, this model meet people to the intuitivism apprehension of color.Using this model, color extraction can be very easily carried out.
It is extracted after license plate range, carries out thick range positioning using empirical equation.Later, in the thick range extracted, one is used
Algorithm of the kind based on Canny operator is further positioned.After the completion of positioning, the present embodiment proposes a kind of based on logo repositioning
Method choose logo range to be sorted, the information of logo itself is not only utilized in this method, also utilize vehicle heat dissipation
The background information of net.It finally carries out gradient orientation histogram (HOG) to extract, be classified by support vector machines (SVM).This implementation
Example using it is a kind of based on feedback by the way of carry out classification results optimization, achieve certain effect.
The foregoing is merely presently preferred embodiments of the present invention, all equivalent changes done according to scope of the present invention patent with
Modification, is all covered by the present invention.
Claims (4)
1. a kind of logo detection and recognition methods based on logo repositioning, it is characterised in that: the following steps are included:
Step S1: the direct picture of variety classes vehicle is obtained, and the direct picture is renamed to mark vehicle
The direct picture is divided into two parts by classification in proportion, wherein 2/3 is used as training set, residue 1/3 is used as test set;
Step S2: license plate area present in the test set direct picture is extracted using hsv color space law, and utilizes form
Learn transformation algorithm detection screening license plate area;
Step S3: the license plate area filtered out by step S2 determines logo range, is then obtained again using the method for repositioning
The car mark region of positioning;
Step S4: Training Support Vector Machines extract histograms of oriented gradients to the car mark region of repositioning, with trained support
Vector machine is classified.
2. a kind of logo detection and recognition methods based on logo repositioning according to claim 1, it is characterised in that: institute
State step S2 specifically includes the following steps:
Step S21: the color space of the direct picture is passed through in openCV image vision library by RGB color
CvtColor function is converted to hsv color space;
Step S22: search step S21 is converted to each of the image in hsv color space pixel, if current search picture
Three channels of vegetarian refreshments meet respectively: the i.e. H range in form and aspect channel between 90-150, purity channel, that is, S range 55-255 it
Between, current pixel point value is then 1, is otherwise just 0, finally obtains a pixel value by lightness channel, that is, V between 30-255
Only 0 and 1 black white image;
Step S23: opening operation is carried out to the black white image that step S22 is obtained using the operator that size is [4,4], can remove and make an uproar
Sound;
Step S24: closed operation, Ke Yilian are carried out to the step S23 image for having removed noise using the operator that size is [20,2]
The region not being filled is connect, a possibility that there are license plates region is obtained;
Step S25: carrying out opening operation using the license plate possibility image that the operator that size is [1,5] obtains step S24, can be with
The edge in the license plate possibility region that removal S24 is obtained is short-tempered;
Step S26: the black white image obtained by step S23, step S24, the complete S22 of step S25 step process finally obtains each
A connected region, i.e. license plate possibility region calculate the ratio of width to height and area of each connected region;According to formula:
Wherein WHi is the length-width ratio of i-th of connected region, and Area is the area of the connected region;Meet above-mentioned formula then to determine
For license plate area.
3. a kind of logo detection and recognition methods based on logo repositioning according to claim 1, it is characterised in that: institute
State step S3 specifically includes the following steps:
Step S31: coarse positioning is carried out to logo, positioning relation is obtained with following formula:
Wherein, lb, rb, tb, bb respectively correspond the left and right up-and-down boundary of the thick range of logo;Lp, rp, bp, tp respectively correspond license plate
Left and right up-and-down boundary;
Step S32: the thick range of logo is handled using Canny edge detection algorithm, obtains the thick volume edges image of logo;
Step S33: the thick volume edges image of logo is obtained to S32 using the operator that size is [1,2] and carries out opening operation, to make
The pixel at logo edge remains;
Step S34: in the logo edge remained due to step S33, some pixels not belong to logo, use statistics picture
Other pixel point methods of plain neighborhood determine whether the pixel that step S33 is remained is logo pixel;Count each picture
Whether the reservation number of pixels in the neighborhood of vegetarian refreshments 7*7 is greater than 30;If number is considered as noise pixel less than 30, by it
Give up;Otherwise retain;
Step S35: the step S34 pixel finally remained is put into pixel home position, is finally retained
Image carries out closed operation using the operator that size is [15,1] to the image finally retained, merges what step S34 was remained
Pixel;
Step S36: repositioning car mark region, and formula is as follows:
Wherein, nlb, nrb, ntb, nbb respectively correspond the left and right up-and-down boundary of logo repositioning range, lp, rp, bp, tp difference
The left and right up-and-down boundary of corresponding license plate.
4. a kind of logo detection and recognition methods based on logo repositioning according to claim 1, it is characterised in that: institute
State step S4 specifically includes the following steps:
Step S41: size normalization is carried out to the logo range of repositioning, is uniformly scaled 64*64;
Step S42: the HOGDescriptor class in openCV image vision library is used normalized repositioning logo range image
Extract histograms of oriented gradients;Wherein block size is 16*16, and cell element size is 8*8, and it is 8*8, gradient direction number that block, which slides increment,
It is 9, the characteristic dimension finally extracted is 1764 dimensions;
Step S43: the training sample image in the training set described in step S1 carries out the detection and extraction of logo, will correctly extract
Logo filter out, be labeled as generic, concurrently set the negative class of a non-logo type;Vector machine type is C class branch
Vector machine is held, sets kernel function as POLY function;With the logo training classifier marked;
Step S44: classification is re-started with trained classifier training set described in step S1, by the non-of erroneous detection positive sample
Logo image tagged is negative class, is fed back to classifier and re-starts training, to enable classifying quality to be promoted.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910331226.2A CN110032991B (en) | 2019-04-23 | 2019-04-23 | Car logo detection and identification method based on car logo relocation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910331226.2A CN110032991B (en) | 2019-04-23 | 2019-04-23 | Car logo detection and identification method based on car logo relocation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110032991A true CN110032991A (en) | 2019-07-19 |
CN110032991B CN110032991B (en) | 2022-07-08 |
Family
ID=67239950
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910331226.2A Active CN110032991B (en) | 2019-04-23 | 2019-04-23 | Car logo detection and identification method based on car logo relocation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110032991B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
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 |
CN106503710A (en) * | 2016-10-26 | 2017-03-15 | 北京邮电大学 | A kind of automobile logo identification method and device |
CN106529424A (en) * | 2016-10-20 | 2017-03-22 | 中山大学 | Vehicle logo recognition method and system based on selective search algorithm |
CN107590492A (en) * | 2017-08-28 | 2018-01-16 | 浙江工业大学 | A kind of vehicle-logo location and recognition methods based on convolutional neural networks |
-
2019
- 2019-04-23 CN CN201910331226.2A patent/CN110032991B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
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 |
CN106529424A (en) * | 2016-10-20 | 2017-03-22 | 中山大学 | Vehicle logo recognition method and system based on selective search algorithm |
US20180114337A1 (en) * | 2016-10-20 | 2018-04-26 | Sun Yat-Sen University | Method and system of detecting and recognizing a vehicle logo based on selective search |
CN106503710A (en) * | 2016-10-26 | 2017-03-15 | 北京邮电大学 | A kind of automobile logo identification method and device |
CN107590492A (en) * | 2017-08-28 | 2018-01-16 | 浙江工业大学 | A kind of vehicle-logo location and recognition methods based on convolutional neural networks |
Non-Patent Citations (3)
Title |
---|
KE,XIAO ET AL.: "Vehicle Logo Recognition with Small Sample Problem in Complex Scene Based on Data Augmentation", 《MATHEMATICAL PROBLEMS IN ENGINEERING》 * |
S. GOPINATHAN ET AL.: "VEHICLE LOGO RECOGNITION USING ENHANCED LEARNING : VLR Recognition", 《PROCEEDINGS OF THE SECOND INTERNATIONAL CONFERENCE ON I-SMAC》 * |
莫文英: "基于图像处理与机器学习的车标及安全带识别研究", 《中国优秀博硕士学位论文全文数据库(硕士)信息科技辑》 * |
Also Published As
Publication number | Publication date |
---|---|
CN110032991B (en) | 2022-07-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103605977B (en) | Extracting method of lane line and device thereof | |
CN105046196B (en) | Front truck information of vehicles structuring output method based on concatenated convolutional neutral net | |
CN102375982B (en) | Multi-character characteristic fused license plate positioning method | |
CN105825203B (en) | Based on point to matching and the matched ground arrow mark detection of geometry and recognition methods | |
Xu et al. | Detection of sudden pedestrian crossings for driving assistance systems | |
CN103871079B (en) | Wireless vehicle tracking based on machine learning and light stream | |
CN101872416B (en) | Vehicle license plate recognition method and system of road image | |
CN103198315B (en) | Based on the Character Segmentation of License Plate of character outline and template matches | |
CN103971097B (en) | Vehicle license plate recognition method and system based on multiscale stroke models | |
CN102867417B (en) | Taxi anti-forgery system and taxi anti-forgery method | |
CN107016362B (en) | Vehicle weight recognition method and system based on vehicle front windshield pasted mark | |
CN102799859B (en) | Method for identifying traffic sign | |
CN105160691A (en) | Color histogram based vehicle body color identification method | |
CN102968637A (en) | Complicated background image and character division method | |
CN103544484A (en) | Traffic sign identification method and system based on SURF | |
CN102902957A (en) | Video-stream-based automatic license plate recognition method | |
CN108205649B (en) | Method and device for recognizing state of driver for calling and answering | |
CN103324958B (en) | Based on the license plate locating method of sciagraphy and SVM under a kind of complex background | |
CN103390167A (en) | Multi-characteristic layered traffic sign identification method | |
CN104463134A (en) | License plate detection method and system | |
WO2024037408A1 (en) | Underground coal mine pedestrian detection method based on image fusion and feature enhancement | |
CN105139011A (en) | Method and apparatus for identifying vehicle based on identification marker image | |
CN105404859A (en) | Vehicle type recognition method based on pooling vehicle image original features | |
CN109145746B (en) | Signal lamp detection method based on image processing | |
Yao et al. | Coupled multivehicle detection and classification with prior objectness measure |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |