CN107330365A - Traffic sign recognition method based on maximum stable extremal region and SVM - Google Patents
Traffic sign recognition method based on maximum stable extremal region and SVM Download PDFInfo
- Publication number
- CN107330365A CN107330365A CN201710388323.6A CN201710388323A CN107330365A CN 107330365 A CN107330365 A CN 107330365A CN 201710388323 A CN201710388323 A CN 201710388323A CN 107330365 A CN107330365 A CN 107330365A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- msup
- image
- mfrac
- 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
- G06V20/582—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of traffic signs
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
- G06T2207/30256—Lane; Road marking
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of traffic sign recognition method based on maximum stable extremal region and SVM, the traffic sign part in RGB image is detected using maximum stable extremal region (MSER) algorithm, gray processing processing, the prominent stability for having played maximum stable extremal region algorithm are carried out to image.The method split with image is detected as edges of regions to be identified with HOG characteristic vectors, the influence that translation and rotating band come can be suppressed to a certain extent, because this method is insensitive for the change of illumination, therefore image can also be reduced because intensity of illumination changes the interference brought.Use SVM classifier in the Classification and Identification stage, it is to avoid fallibility and the machine training of handmarking it is a large amount of time-consuming, this method preferably balances the requirement of accuracy and real-time, realize automatic detection and the identification of traffic sign.The invention is recognized to the test pictures in German traffic mark examination criteria (German Traffic Sign Detection Benchmark) database, has obtained better effects.
Description
Technical field
The invention belongs to image processing field, apply in intelligent transportation scene, be related to and apply maximum stable extremal region
(MSER) algorithm, realizes effective processing to traffic sign part in RGB image, application direction histogram of gradients (HOG) is to emerging
Interesting region carries out edge detection and segmentation, finally sends into vector machine classifier (SVM) to complete the work of cognitive phase.
Background technology
Image processing techniques is important part in intelligent transportation field, and the automation traffic sign of efficiently and accurately is known
Traffic behavior is participated in while traffic participant specification can not guided, mitigates driver information processing pressure, so as to reduce accident hair
Raw probability.At present, traffic mark identification (Traffic Sign Recognition, TSR) system is mainly by installed in car
The traffic mark information on camera acquisition road on, is sent to image processing module and carries out Mark Detection and identification, it
System will make different counter-measures according to the result of identification afterwards.Detection-phase needs the color and shape according to traffic mark
Feature, finds and positions the area-of-interest for including traffic mark from the image of collection.Cognitive phase will to area-of-interest,
Feature is extracted with different methods, and these area-of-interests are classified with suitable sorting algorithm, traffic mark is obtained
Type information.
With going deep into for research, the theoretical method of many automatic recognition of traffic signs is emerged.Current most identification
The theoretical method all split in detection-phase using colouring information as image, still, colouring information by strong light, dim light,
Effect can be deteriorated under the influence of situations such as haze, fog, mark self color degrade.Although in addition, traffic mark generally has
Eye-catching color, naked eyes are readily discernible, but under the City scenarios of modernization (such as colorful light, wall color etc.), make
Obtain traffic mark to be difficult to differentiate between with background color, so as to be difficult to find identified areas during Mark Detection.In cognitive phase,
The grader that most of existing system is trained by the true picture using manual markings is constituted, and this is that a repetition takes, and
And the processing procedure easily malfunctioned.Therefore, to evade this artificial operation and manual markings training as far as possible.Simultaneously as traffic
Sign recognition system is higher to the requirement of real-time of detection identification, especially in the case where condition of road surface complexity, speed are higher, is
System is needed to make identification within the time as short as possible, and result is notified into driver, so that the time that driver has abundance does
Go out reaction.Therefore, in view of colour recognition is the drawbacks of complex environment is applied, the training of the fallibility and machine of manual markings it is a large amount of
It is time-consuming, and requirement of the system to real-time, effect more stable algorithm more balanced without a kind of performance, to realize friendship
The Intelligent Recognition of logical mark.
The content of the invention
Currently invention addresses several important performance indexes in Traffic Sign Recognition, it is proposed that one kind is based on maximum stable pole
It is worth region and SVM traffic sign recognition method, it is therefore intended that the performance of optimization prior art method, overcomes the shortcomings of prior art.
It is main that extraction segmentation is carried out to the picture containing traffic sign with methods such as maximum stable extremal regions, will be initial common
RGB photos, pre-process into the bianry image beneficial to recognition detection；Feature extraction is carried out using HOG and send into SVM classifier afterwards
It is middle to be classified and recognized.
Technical scheme：
Traffic sign recognition method based on maximum stable extremal region and SVM, step is as follows：
Step 1. color is changed：Interest region is detected using maximum stable extremal region (MSER) algorithm, and it is right
Image carries out gray processing processing.
Step 2. rim detection：The horizontal and vertical First-order Gradient of each pixel in calculating interest region, and as
HOG characteristic vectors carry out rim detection to interest region.
Step 3. Classification and Identification：The shape in interest region is classified and recognized with SVM classifier.And according to data
Standard picture in storehouse, exports recognized traffic sign.
Beneficial effects of the present invention：Traffic sign based on maximum stable extremal region and SVM proposed by the invention is known
Other method, the prominent stability for having played maximum stable extremal region algorithm, and it is used as interest regional edge with HOG characteristic vectors
The method that edge is detected and image is split, this method can suppress the influence that translation and rotating band come to a certain extent, due to it
Change for illumination is insensitive, therefore can also reduce image because intensity of illumination changes the interference brought, in the Classification and Identification stage
Use SVM classifier, it is to avoid a large amount of time-consuming, the advantages of comprehensive each method of fallibility and the machine training of handmarking, compared with
The requirement of accuracy and real-time is balanced well, realizes automatic detection and the identification of traffic sign.
Brief description of the drawings
Fig. 1 is the flow chart of traffic sign recognition method of the present invention
Fig. 2 is the design sketch of the example recognition flow of the present invention
Embodiment
Below in conjunction with accompanying drawing and technical scheme, the embodiment of the present invention is further illustrated.
For above-mentioned 3 steps, detailed description below is carried out to each step：
Step 1：Color is changed
Image normalization：Because traffic sign is generally red and blueness, first formula
Red and blue portion is counted, larger one is chosen as threshold value, red blue normalization is carried out to picture
Processing.
Step 2：Rim detection
2.1 calculate image gradient.In image, neighbor pixel gray-value variation is less, and its gradient magnitude is also smaller；
Conversely, the place of gray scale mutation, gradient magnitude is larger.The size of amplitude is obtained come in detail accordingly, it would be desirable to be calculated by single order inverse
The position of thin process decision chart picture and the existence at edge.And can be by the calculating of second dervative for different edge gray values
To determine, the position at edge is exactly the zero crossing of second dervative, and its gradient is the first derivative of corresponding image function.In image
(x, y) point is selected in function f (x, y), its correspondence gradient is：
Wherein Gx(x, y) and Gy(x, y) is the horizontal direction gradient and vertical gradient at pixel (x, y) place respectively.f
(x, y) value that unit length changes on its maximum rate of change direction is the amplitude of correspondence gradient.Corresponding gradient magnitude
And the relationship at respective direction angle is:
Replace differential with difference to realize the calculating of gradient for image function f (x, y), i.e.,：
Therefore, the gradient magnitude G (x, y) and gradient direction α (x, y) at pixel (x, y) place can be expressed as：
2.2 be that each cell factory builds gradient orientation histogram.After a gradient image is obtained, image is cut into
The grid of formed objects (8x8pixel) is referred to as cell, to the direction inside each cell statistics, tan-1Angular range be situated between
Arrived in -90 degree between+90, its angle is moved to 0 and arrives a80 degree, and is evenly dividing as the ballot box of 9 pieces, i.e., nine, phase between every piece
Poor 20 degree, when the direction of pixel meets some ballot box, just the intensity level (Magnitude) of the pixel is put into relatively to cumulative
In the ballot box answered, each cell can be represented with nine dimensional vectors, can finally draw this cell histograms of oriented gradients.
2.3 cell factories are combined into normalized gradient histogram in big block (block), block.Per 4 adjacent cell
A block (block) is constituted, the characteristic vector connection in a block is got up to obtain the characteristic vector of 36 dimensions, block is used to sample image
It is scanned, scanning step is a cell.Characteristic vector quantity is represented with N：
Wherein B=4, H=9, R are interest regions, and for the image of a width 64*64, common property gives birth to the feature of 1764 dimensions
The HOG features of vector, the i.e. region.
Step 3. Classification and Identification
The structure choice of 3.1 Multi- class SVM classifiers.SVM is a kind of binary classifier, in order to be generalized to multiclass point
Class, the strategy that the present invention takes is a series of two classes SVM classifiers of construction, and each grader is used to recognize two of which classification,
And their differentiation result is combined in some way realize it is more than two classes and two classes classify.In order to realize multicategory discriminant,
It is 1-aginst-rest algorithms with 1-a-r, N number of binary classifier, the i-th classes of i-th of SVM is constructed for N number of classification problem
In training sample as positive training sample, and using other samples as negative training sample, finally output is two classes point
Class device is output as that class of maximum, and this method is simple, easily realizes.
The training and identification of 3.2SVM graders.With standard traffic marking pattern image set, SVM classifier is trained, obtained
To the corresponding decision function of supporting vector and supporting vector and class label of training sample.By the image by pretreatment
Characteristic vector inputs SVM classifier, judges and output category result.
Claims (4)
1. the traffic sign recognition method based on maximum stable extremal region and SVM, this method comprises the following steps：
Step 1：Color is changed.
Step 2：Rim detection.
Step 3：Classification and Identification.
2. the method in claim 1 described in step 1, is characterised by, the step 1 includes：
Step 1a：Image red blue normalized threshold is determined, calculation formula is：
<mrow>
<msub>
<mi>&Omega;</mi>
<mrow>
<mi>R</mi>
<mi>B</mi>
</mrow>
</msub>
<mo>=</mo>
<mi>m</mi>
<mi>a</mi>
<mi>x</mi>
<mrow>
<mo>(</mo>
<mfrac>
<mi>R</mi>
<mrow>
<mi>R</mi>
<mo>+</mo>
<mi>G</mi>
<mo>+</mo>
<mi>B</mi>
</mrow>
</mfrac>
<mo>,</mo>
<mfrac>
<mi>B</mi>
<mrow>
<mi>R</mi>
<mo>+</mo>
<mi>G</mi>
<mo>+</mo>
<mi>B</mi>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
</mrow>
Wherein R is RED sector, and B is blue portion, and R+G+B is that image is whole.ΩRBIt is exactly the normalized threshold value of red blue.
Step 1b：Using the threshold value calculated, image is normalized, the interest region of prominent identification to be detected.
3. step 2 methods described in claim 1, it is characterised in that the step 2 includes：
2a：Calculate image gradient.In image, neighbor pixel gray-value variation is less, and its gradient magnitude is also smaller；Instead
It, the place of gray scale mutation, gradient magnitude is larger.The size of amplitude is obtained come in detail accordingly, it would be desirable to be calculated by single order inverse
The position of process decision chart picture and the existence at edge.(x, y) point is selected in image function f (x, y), its correspondence gradient formula
For：
<mrow>
<mo>&dtri;</mo>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mrow>
<mo>&lsqb;</mo>
<msub>
<mi>G</mi>
<mi>x</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<msub>
<mi>G</mi>
<mi>y</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>&rsqb;</mo>
</mrow>
<mi>T</mi>
</msup>
<mo>=</mo>
<msup>
<mrow>
<mo>&lsqb;</mo>
<mfrac>
<mrow>
<mo>&part;</mo>
<mi>f</mi>
</mrow>
<mrow>
<mo>&part;</mo>
<mi>x</mi>
</mrow>
</mfrac>
<mfrac>
<mrow>
<mo>&part;</mo>
<mi>f</mi>
</mrow>
<mrow>
<mo>&part;</mo>
<mi>y</mi>
</mrow>
</mfrac>
<mo>&rsqb;</mo>
</mrow>
<mi>T</mi>
</msup>
</mrow>
Wherein Gx(x, y) and Gy(x, y) is the horizontal direction gradient and vertical gradient at pixel (x, y) place respectively.F (x, y)
The value that unit length changes on its maximum rate of change direction is the amplitude of correspondence gradient.Corresponding gradient magnitude and phase
The relationship for answering deflection is:
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mo>|</mo>
<mo>&dtri;</mo>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>|</mo>
<mo>=</mo>
<mi>m</mi>
<mi>a</mi>
<mi>g</mi>
<mrow>
<mo>(</mo>
<mo>&dtri;</mo>
<mi>f</mi>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msqrt>
<mrow>
<msub>
<mi>G</mi>
<mi>x</mi>
</msub>
<msup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>+</mo>
<msub>
<mi>G</mi>
<mi>y</mi>
</msub>
<msup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</msqrt>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>&Phi;</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mi>tan</mi>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msup>
<mfrac>
<mrow>
<msub>
<mi>G</mi>
<mi>y</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msub>
<mi>G</mi>
<mi>x</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
Replace differential with difference to realize the calculating of gradient for image function f (x, y), i.e.,：
<mrow>
<mo>|</mo>
<mo>&dtri;</mo>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>|</mo>
<mo>=</mo>
<msqrt>
<mrow>
<mo>{</mo>
<msup>
<mrow>
<mo>&lsqb;</mo>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>+</mo>
<mn>1</mn>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>&rsqb;</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>+</mo>
<msup>
<mrow>
<mo>&lsqb;</mo>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>+</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mo>&rsqb;</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>}</mo>
</mrow>
</msqrt>
</mrow>
Therefore, the gradient magnitude G (x, y) and gradient direction α (x, y) at pixel (x, y) place can be expressed as：
<mrow>
<mi>G</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msqrt>
<mrow>
<msub>
<mi>G</mi>
<mi>x</mi>
</msub>
<msup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>+</mo>
<msub>
<mi>G</mi>
<mi>y</mi>
</msub>
<msup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</msqrt>
</mrow>
<mrow>
<mi>&alpha;</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mi>tan</mi>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msup>
<mo>&lsqb;</mo>
<mfrac>
<mrow>
<msub>
<mi>G</mi>
<mi>y</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msub>
<mi>G</mi>
<mi>x</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>&rsqb;</mo>
</mrow>
2b：Gradient orientation histogram is built for each cell factory.After a gradient image is obtained, image is cut into identical
The grid of size (8x8pixel) is referred to as cell, to the direction inside each cell statistics, tan-1Angular range be between -90
Spend between+90, its angle is moved to 0 and arrives t80 degree, and be evenly dividing as the ballot box of 9 pieces, i.e., nine, 20 are differed between every piece
Degree, when the direction of pixel meets some ballot box, is just put into the intensity level (Magnitude) of the pixel to adding up corresponding
In ballot box, each cell can be represented with nine dimensional vectors, can finally draw this cell histograms of oriented gradients.
2c：Cell factory is combined into normalized gradient histogram in big block (block), block.Constituted per 4 adjacent cell
One block (block), gets up to obtain the characteristic vector of 36 dimensions, sample image is carried out with block the characteristic vector connection in a block
Scanning, scanning step is a cell.Characteristic vector quantity is represented with N：
<mrow>
<mi>N</mi>
<mo>=</mo>
<mrow>
<mo>(</mo>
<mfrac>
<msub>
<mi>R</mi>
<mrow>
<mi>w</mi>
<mi>i</mi>
<mi>d</mi>
<mi>t</mi>
<mi>h</mi>
</mrow>
</msub>
<msub>
<mi>M</mi>
<mrow>
<mi>w</mi>
<mi>i</mi>
<mi>d</mi>
<mi>t</mi>
<mi>h</mi>
</mrow>
</msub>
</mfrac>
<mo>-</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mo>*</mo>
<mrow>
<mo>(</mo>
<mfrac>
<msub>
<mi>R</mi>
<mrow>
<mi>h</mi>
<mi>e</mi>
<mi>i</mi>
<mi>g</mi>
<mi>h</mi>
<mi>t</mi>
</mrow>
</msub>
<msub>
<mi>M</mi>
<mrow>
<mi>h</mi>
<mi>e</mi>
<mi>i</mi>
<mi>g</mi>
<mi>h</mi>
<mi>t</mi>
</mrow>
</msub>
</mfrac>
<mo>-</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mo>*</mo>
<mi>B</mi>
<mo>*</mo>
<mi>H</mi>
</mrow>
4. according to step 3 methods described in claim 1, being characterised by, the step 3 includes：
3a：The structure choice of Multi- class SVM classifier.SVM is a kind of binary classifier, in order to be generalized to multicategory classification, this
It is a series of two classes SVM classifiers of construction to invent the strategy taken, and each grader is used to recognizing two of which classification, and by it
Differentiation result combine in some way realize it is more than two classes and two classes classify.In order to realize multicategory discriminant, with 1-
A-r is 1-aginst-rest algorithms, and N number of binary classifier, instructions of i-th of SVM in the i-th class are constructed for N number of classification problem
Practice sample as positive training sample, and using other samples as negative training sample, finally output is that binary classifier is defeated
Go out for that maximum class, this method is simple, easily realizes.
3b：The training and identification of SVM classifier.With standard traffic marking pattern image set, SVM classifier is trained, instructed
Practice the supporting vector and the corresponding decision function of supporting vector and class label of sample.By the feature of the image by pretreatment
Vector input SVM classifier, judges and output category result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710388323.6A CN107330365A (en) | 2017-05-27 | 2017-05-27 | Traffic sign recognition method based on maximum stable extremal region and SVM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710388323.6A CN107330365A (en) | 2017-05-27 | 2017-05-27 | Traffic sign recognition method based on maximum stable extremal region and SVM |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107330365A true CN107330365A (en) | 2017-11-07 |
Family
ID=60193099
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710388323.6A Pending CN107330365A (en) | 2017-05-27 | 2017-05-27 | Traffic sign recognition method based on maximum stable extremal region and SVM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107330365A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108154102A (en) * | 2017-12-21 | 2018-06-12 | 安徽师范大学 | A kind of traffic sign recognition method |
| CN108182431A (en) * | 2018-03-15 | 2018-06-19 | 大连理工大学 | A kind of traffic sign recognition method based on maximum stable extremal region and genetic optimization SVM |
| CN109063619A (en) * | 2018-07-25 | 2018-12-21 | 东北大学 | A kind of traffic lights detection method and system based on adaptive background suppression filter and combinations of directions histogram of gradients |
| CN109919182A (en) * | 2019-01-24 | 2019-06-21 | 国网浙江省电力有限公司电力科学研究院 | A kind of terminal side electric power safety operation image-recognizing method |
| CN113743351A (en) * | 2021-09-14 | 2021-12-03 | 北京石油化工学院 | Remote sensing image scene recognition method based on edge direction semantic information |
| CN113839930A (en) * | 2021-09-06 | 2021-12-24 | 哈尔滨工业大学 | Network intrusion detection method and system based on image processing |
| CN115100228A (en) * | 2022-07-25 | 2022-09-23 | 江西现代职业技术学院 | Image processing method, system, readable storage medium and computer device |
| CN116311212A (en) * | 2023-05-15 | 2023-06-23 | 青岛恒天翼信息科技有限公司 | Ship number identification method and device based on high-speed camera and in motion state |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104809437A (en) * | 2015-04-28 | 2015-07-29 | 无锡赛睿科技有限公司 | Real-time video based vehicle detecting and tracking method |
| CN104866862A (en) * | 2015-04-27 | 2015-08-26 | 中南大学 | Strip steel surface area type defect identification and classification method |
| CN105787475A (en) * | 2016-03-29 | 2016-07-20 | 西南交通大学 | Traffic sign detection and identification method under complex environment |
| US20170323452A1 (en) * | 2014-11-18 | 2017-11-09 | Agency For Science, Technology And Research | Method and device for traffic sign recognition |
-
2017
- 2017-05-27 CN CN201710388323.6A patent/CN107330365A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170323452A1 (en) * | 2014-11-18 | 2017-11-09 | Agency For Science, Technology And Research | Method and device for traffic sign recognition |
| CN104866862A (en) * | 2015-04-27 | 2015-08-26 | 中南大学 | Strip steel surface area type defect identification and classification method |
| CN104809437A (en) * | 2015-04-28 | 2015-07-29 | 无锡赛睿科技有限公司 | Real-time video based vehicle detecting and tracking method |
| CN105787475A (en) * | 2016-03-29 | 2016-07-20 | 西南交通大学 | Traffic sign detection and identification method under complex environment |
Non-Patent Citations (2)
| Title |
|---|
| 叶阳阳: "交通标志检测和识别算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 鲁寒凝: "基于HOG特征的交通标志检测与识别算法研究", 《中国优秀硕士论文全文数据库 信息科技辑》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108154102A (en) * | 2017-12-21 | 2018-06-12 | 安徽师范大学 | A kind of traffic sign recognition method |
| CN108154102B (en) * | 2017-12-21 | 2021-12-10 | 安徽师范大学 | Road traffic sign identification method |
| CN108182431A (en) * | 2018-03-15 | 2018-06-19 | 大连理工大学 | A kind of traffic sign recognition method based on maximum stable extremal region and genetic optimization SVM |
| CN109063619A (en) * | 2018-07-25 | 2018-12-21 | 东北大学 | A kind of traffic lights detection method and system based on adaptive background suppression filter and combinations of directions histogram of gradients |
| CN109919182A (en) * | 2019-01-24 | 2019-06-21 | 国网浙江省电力有限公司电力科学研究院 | A kind of terminal side electric power safety operation image-recognizing method |
| CN113839930A (en) * | 2021-09-06 | 2021-12-24 | 哈尔滨工业大学 | Network intrusion detection method and system based on image processing |
| CN113743351A (en) * | 2021-09-14 | 2021-12-03 | 北京石油化工学院 | Remote sensing image scene recognition method based on edge direction semantic information |
| CN113743351B (en) * | 2021-09-14 | 2023-07-04 | 北京石油化工学院 | Remote sensing image scene recognition method based on edge direction semantic information |
| CN115100228A (en) * | 2022-07-25 | 2022-09-23 | 江西现代职业技术学院 | Image processing method, system, readable storage medium and computer device |
| CN116311212A (en) * | 2023-05-15 | 2023-06-23 | 青岛恒天翼信息科技有限公司 | Ship number identification method and device based on high-speed camera and in motion state |
| CN116311212B (en) * | 2023-05-15 | 2023-08-18 | 青岛恒天翼信息科技有限公司 | Ship number identification method and device based on high-speed camera and in motion state |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107330365A (en) | Traffic sign recognition method based on maximum stable extremal region and SVM | |
| CN105261017B (en) | The method that image segmentation based on road surface constraint extracts pedestrian's area-of-interest | |
| CN104599275B (en) | The RGB-D scene understanding methods of imparametrization based on probability graph model | |
| Ruta et al. | Real-time traffic sign recognition from video by class-specific discriminative features | |
| CN103366190B (en) | A kind of method of identification traffic signss | |
| CN109101924A (en) | A kind of pavement marking recognition methods based on machine learning | |
| CN114067143B (en) | Vehicle re-identification method based on double sub-networks | |
| Le et al. | Real time traffic sign detection using color and shape-based features | |
| CN112016605B (en) | Target detection method based on corner alignment and boundary matching of bounding box | |
| CN105488536A (en) | Agricultural pest image recognition method based on multi-feature deep learning technology | |
| CN105005989B (en) | A kind of vehicle target dividing method under weak contrast | |
| CN106529532A (en) | License plate identification system based on integral feature channels and gray projection | |
| CN106709530A (en) | License plate recognition method based on video | |
| CN1996346A (en) | Vehicle recognition algorithm based on contour | |
| CN108090459B (en) | Traffic sign detection and identification method suitable for vehicle-mounted vision system | |
| CN109697397B (en) | Target detection method and device | |
| CN102968646A (en) | Plate number detecting method based on machine learning | |
| CN105303169B (en) | A kind of cell division identification method and its identification device based on slow feature | |
| CN108090928B (en) | Method and system for detecting and screening quasi-circular cell area | |
| CN104200228A (en) | Recognizing method and system for safety belt | |
| CN104050684A (en) | Video moving object classification method and system based on on-line training | |
| CN108073940B (en) | Method for detecting 3D target example object in unstructured environment | |
| CN106372658A (en) | Vehicle classifier training method | |
| CN107315990A (en) | A kind of pedestrian detection algorithm based on XCS LBP features and cascade AKSVM | |
| Xing et al. | Traffic sign detection and recognition using color standardization and Zernike moments |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20171107 |