CN110226170A - A kind of traffic sign recognition method in rain and snow weather - Google Patents
A kind of traffic sign recognition method in rain and snow weather Download PDFInfo
- Publication number
- CN110226170A CN110226170A CN201880005497.2A CN201880005497A CN110226170A CN 110226170 A CN110226170 A CN 110226170A CN 201880005497 A CN201880005497 A CN 201880005497A CN 110226170 A CN110226170 A CN 110226170A
- Authority
- CN
- China
- Prior art keywords
- convolutional neural
- neural networks
- image
- sign board
- sign
- 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
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 54
- 238000001514 detection method Methods 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims description 51
- 238000012545 processing Methods 0.000 claims description 10
- 238000001914 filtration Methods 0.000 claims description 7
- 238000011478 gradient descent method Methods 0.000 claims description 6
- 230000000644 propagated effect Effects 0.000 claims description 6
- 230000015556 catabolic process Effects 0.000 claims description 4
- 238000004040 coloring Methods 0.000 claims description 4
- 230000000877 morphologic effect Effects 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 230000001537 neural effect Effects 0.000 description 4
- 230000003044 adaptive effect Effects 0.000 description 2
- 230000002411 adverse Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 239000003086 colorant Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000007850 degeneration Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000003595 mist Substances 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- 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
-
- 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
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
Abstract
A kind of traffic sign recognition method in rain and snow weather, comprising: the image of acquisition is handled, the clear image (S11) after obtaining sleet;Sign board detection (S12) is carried out to the clear image;Board type identification (S13) is identified to the Sign Board image detected.Type identification is carried out to Sign Board using tandem type convolutional neural networks, improves the efficiency and accuracy of classification.Road signs in rain and snow weather can quickly be accurately identified, it advantageously accounts under bad weather, driver due to sight is hindered and the problem of can not timely and accurately capture the flag information in road, advantageously ensure that traffic safety and improve conevying efficiency.
Description
Technical field
The present invention relates to field of intelligent transportation technology, and in particular to the traffic sign recognition side in a kind of rain and snow weather
Method.
Background technique
As the fast development of science and technology and the whole of people's living standard improve, China's car ownership significantly increases in recent years
Add.Automobile bring trip it is convenient while, also bring obvious pressure to urban transportation, cause traffic safety
Problem and conevying efficiency problem become to become increasingly conspicuous.Traffic sign recognition (Traffic Sign Recognition, letter
Claim TSR) branch field as intelligent transportation system, pass through the knowledge that detected and classified to the road signs in scene
Not, the key message of road traffic instruction is obtained, it has also become the hot spot of intelligent transportation research.
In the prior art for the research of Traffic Sign Recognition problem mainly for good weather condition, and to adverse weather
If the Study of recognition in the case of haze, rain, snow is less, this has resulted in having the limitation of TSR system in practical applications at present
Property.Under the bad weathers such as mist, rain, snow, since the sight of driver is hindered, the safe visual field can become very narrow, sight distance
From shortening, the flag information in road can not be timely and accurately captured, is hidden some dangers for for traffic safety.In vehicle driving
In, especially under poor weather, driver often more needs reliable traffic sign recognition method that it is assisted to drive
It sails.In addition, reliably obtaining road signs information in adverse weather is also the technology that unmanned field needs to solve
One of problem.
Summary of the invention
In view of this, it is an object of the invention to overcome the deficiencies of the prior art and provide the roads in a kind of rain and snow weather
Traffic sign recognition method.
In order to achieve the above object, the present invention adopts the following technical scheme: the road signs in a kind of rain and snow weather are known
Other method, comprising:
The image of acquisition is handled, the clear image after obtaining sleet;
Sign board detection is carried out to the clear image;
Board type identification is identified to the Sign Board image detected.
Optionally, the image of described pair of acquisition is handled, comprising:
It is coarse figure and detail view by the picture breakdown of acquisition;
The rainprint in detail view is removed, to obtain clear image.
Optionally, the picture breakdown by acquisition is coarse figure and detail view, comprising:
The image of acquisition is subjected to low-pass filtering treatment, the image obtained after processing is coarse figure;
The image of acquisition and coarse figure are subtracted each other, detail view is obtained.
Optionally, the sleet trace in the removal detail view includes:
Using the morphological difference between rainprint and traffic sign texture, by sparse dictionary learning algorithm by the details
Figure is divided into texture maps and rainprint figure.
Optionally, the sleet trace in the removal detail view further include:
According to the prior information of rainprint shape feature, the texture maps and rainprint figure are carried out by rainprint length-width ratio secondary
Differentiate, is come with more accurately decompositing texture maps from detail view.
It is optionally, described that sign board detection is carried out to the clear image, comprising:
Establish multilayer feature conspicuousness model;
The clear image is inputted into the multilayer feature conspicuousness model, to obtain sign board testing result.
It is optionally, described to establish multilayer feature conspicuousness model, comprising:
The distinctive information of sign board is extracted, is trained by the information input training pattern, and using boosting algorithm.
Optionally, the distinctive information of the sign board includes:
Shape information, colouring information, gradient information and location information.
Optionally, the described pair of Sign Board image detected is identified board type identification, comprising:
Type identification is carried out to Sign Board using tandem type convolutional neural networks;
Wherein, the tandem type convolutional neural networks include: first order convolutional neural networks and second level convolutional Neural net
Network;
The first order convolutional neural networks carry out rough sort to the Sign Board image of input, and the result of rough sort is defeated
Enter to second level convolutional neural networks and be finely divided class, to identify the type of Sign Board.
Optionally, the tandem type convolutional neural networks carry out type identification to Sign Board, and detailed process includes:
Establish training sample data;
Stable first order convolutional neural networks and second level convolutional neural networks are trained according to training sample data;
First order convolutional neural networks and second level convolutional neural networks are subjected to cascade and form tandem type convolutional Neural net
Network;
Sign board type identification is carried out using the tandem type convolutional neural networks that training obtains.
Optionally, the training sample data include:
The Traffic Sign Images that actual acquisition arrives, and
Gaussian noise and the image after rotation and scaling processing are added to the Traffic Sign Images that actual acquisition arrives.
It is optionally, described that stable first order convolutional neural networks are trained according to training sample data, comprising:
Training sample data are subjected to propagated forward and reverse conduction, convolution kernel and biasing are completed by gradient descent method
Update, alternately and repeatedly handle, until meeting the identification condition of convergence or until reach the frequency of training of requirement, training obtain feature to
Amount;
By this feature vector training SVM classifier, first order convolutional neural networks are obtained.
It is optionally, described that stable second level convolutional neural networks are trained according to training sample data, comprising:
Speed(-)limit sign, prohibitory sign, caution sign, round Warning Mark, rectangular finger are picked out in training sample data
The training sample data of indicating will, fingerpost as second level convolutional neural networks;
The training sample data are subjected to propagated forward and reverse conduction, convolution kernel and biasing are completed by gradient descent method
Update, alternately and repeatedly handle, until meeting the identification condition of convergence or until reach the frequency of training of requirement, training obtains feature
Vector;
By this feature vector training SVM classifier, second level convolutional neural networks are obtained.
Optionally, this method further include:
The Sign Board detected is tracked using Kalman filtering and Camshift algorithm.
The invention adopts the above technical scheme, the traffic sign recognition method in the rain and snow weather, comprising: to adopting
The image of collection is handled, the clear image after obtaining sleet;Sign board detection is carried out to the clear image;To inspection
The Sign Board image measured is identified board type identification.Technical solution proposed by the invention can be to the road in rain and snow weather
Road traffic sign is quickly accurately identified, and is advantageously accounted under bad weather, and driver can not since sight is hindered
The problem of timely and accurately capturing the flag information in road advantageously ensures that traffic safety and improves conevying efficiency.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is the flow diagram of traffic sign recognition embodiment of the method one of the present invention;
Fig. 2 is the flow diagram for carrying out sleet processing in the embodiment of the present invention one to the image of acquisition;
Fig. 3 is to carry out sign board detection by image of the multilayer feature conspicuousness model to acquisition in the embodiment of the present invention one
Schematic diagram;
Fig. 4 is the result figure that the present invention carries out sign board detection to the image of acquisition;
Fig. 5 is the structural schematic diagram of the BaseNet model provided in the embodiment of the present invention;
Fig. 6 is that the tandem type convolutional neural networks provided in the embodiment of the present invention carry out the signal of type identification to Sign Board
Figure.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, technical solution of the present invention will be carried out below
Detailed description.Obviously, described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.Base
Embodiment in the present invention, those of ordinary skill in the art are obtained all without making creative work
Other embodiment belongs to the range that the present invention is protected.
As shown in Figure 1, as the embodiment of the present invention one, the traffic sign recognition side in a kind of rain and snow weather is provided
Method, comprising:
S11: handling the image of acquisition, the clear image after obtaining sleet;
S12: sign board detection is carried out to the clear image;
S13: board type identification is identified to the Sign Board image detected.
The sign board color in China is relatively more fixed, mainly has red, yellow, blue three-color, and these colors are in sleet sky
There is certain degeneration in gas.By the different existence forms of analysis sleet point, sleet line in Traffic Sign Images, we will scheme
Linear superposition as being decomposed into coarse figure and detail view, is formulaically expressed as:
J (x)=B (x)+R (x) (1)
Wherein, B (x) is coarse figure, and R (x) is detail view.For the image in sleet scene, detail view by texture layer and
Rainprint layer cooperatively forms, and is formulaically expressed as:
Wherein, T (x) is texture maps, and K (x) is rainprint figure.Therefore, using by coarse figure with go the detail view phase after rainprint
Calais's reconstructed image finally obtains recovery clear image are as follows:
I (x)=B (x)+T (x) (3)
Based on above-mentioned model and thinking, the traffic mark board image of sleet is removed in order to obtain, it is necessary to single width sleet figure
As J (x) is handled to obtain B (x), R (x), T (x) and K (x).
As shown in Fig. 2, the image of acquisition is carried out low-pass filtering treatment, such as using weighted least-squares method filter to defeated
Enter image and carry out edge-smoothing processing, the image obtained after processing is coarse figure B (x);
The image of acquisition and coarse figure are subtracted each other, detail view R (x) is obtained.
Further, the sleet trace in the removal detail view includes:
Using the morphological difference between rainprint and traffic sign texture, by sparse dictionary learning algorithm by the details
Figure is divided into texture maps and rainprint figure.
Specific treatment process may is that
Pass through histograms of oriented gradients feature (HOG, Histogram of Oriented Gradient) and local binary
Mode (LBP, Local Binary Patterns) describes the morphological difference between rain line and texture, and uses for reference based on son
Details subgraph dictionary is divided into texture dictionary and rainprint dictionary using this otherness by the image prior information modeling thought of block,
Finally corresponding sparse coefficient is obtained using orthogonal matching pursuit (OMP, Orthogonal Matching Pursuit) algorithm.
During dictionary classification, we will use the Regularization Strategy based on sparse constraint, by image moderate rain itself
Prior information be introduced into dictionary classification during, the atom of misclassification by rain line aspect ratio carries out sub- dictionary secondary
Differentiate, finally disassembles T (x) and K (x) from R (x) to come, so that reaching image goes rainprint effect.
It is further, described that sign board detection is carried out to the clear image, comprising:
Establish multilayer feature conspicuousness model;
The clear image is inputted into the multilayer feature conspicuousness model, to obtain sign board testing result.
It is further, described to establish multilayer feature conspicuousness model, comprising:
The distinctive information of sign board is extracted, is trained by the information input training pattern, and using boosting algorithm.
The real-time that will affect actually detected system if the feature set number extracted is more, needs using effective machine
System extracts most effective feature from obtained feature set.The choosing of optimal characteristics is realized in the present embodiment using boosting algorithm
It selects, to obtain most effective feature.As shown in figure 3, the multilayer feature conspicuousness model can be the vision based on Itti algorithm
Attention model, wherein h is verification and measurement ratio, and f is false alarm rate, and N is series.
It is understood that the distinctive information of sign board includes:
Shape information, i.e. triangle, rectangular and round etc. shape informations;
The colouring informations such as colouring information, i.e. traffic mark board fixed red, blue, white and yellow;
Gradient information is the gradient information of eight different directions in gray space;
Location information is the location information that sign board often occurs in the visual field.
Fig. 4 is the result figure for carrying out sign board detection to the image of acquisition using above method.(b) in Fig. 4, (c),
(d), the extraction result of four kinds of different characteristics (e) is illustrated in four figures.
Further, the described pair of Sign Board image detected is identified board type identification, comprising:
Type identification is carried out to Sign Board using tandem type convolutional neural networks;
Wherein, the tandem type convolutional neural networks include: first order convolutional neural networks and second level convolutional Neural net
Network;
The first order convolutional neural networks carry out rough sort to the Sign Board image of input, and the result of rough sort is defeated
Enter to second level convolutional neural networks and be finely divided class, to identify the type of Sign Board.
The present embodiment carries out type identification to Sign Board by using tandem type convolutional neural networks, is able to ascend classification
Efficiency and accuracy.
Specifically, the tandem type convolutional neural networks carry out type identification to Sign Board, detailed process includes:
Establish training sample data;
Stable first order convolutional neural networks and second level convolutional neural networks are trained according to training sample data;
First order convolutional neural networks and second level convolutional neural networks are subjected to cascade and form tandem type convolutional Neural net
Network;
Sign board type identification is carried out using the tandem type convolutional neural networks that training obtains.
The processing occupancy whole service time specific gravity for considering the convolutional layer in convolutional neural networks operational process is larger, this reality
The structure by simplifying convolutional layer is applied to promote the speed of service of network;Meanwhile in order to overcome the simplification because of convolutional layer to lead to net
Network extracts the problem of characteristic pattern is reduced, and increases the feature of output using the strategy that maximum value sampling and average value sampling are combined
Quantity;And it is used as using SVM (Support Vector Machine, support vector machines) construction optimal separating hyper plane and reaches complete
The optimal classifier of office.Its network structure is as shown in figure 5, be named as BaseNet model.
Further, the training sample data include:
The Traffic Sign Images that actual acquisition arrives, and
Gaussian noise and the image after rotation and scaling processing are added to the Traffic Sign Images that actual acquisition arrives.
It is described that stable first order convolutional neural networks are trained according to training sample data, comprising:
Training sample data are subjected to propagated forward and reverse conduction, convolution kernel and biasing are completed by gradient descent method
Update, alternately and repeatedly handle, until meeting the identification condition of convergence or until reach the frequency of training of requirement, training obtain feature to
Amount;
By this feature vector training SVM classifier, level-one BaseNet model is obtained.
It is further, described that stable second level convolutional neural networks are trained according to training sample data, comprising:
Speed(-)limit sign, prohibitory sign, caution sign, round Warning Mark, rectangular finger are picked out in training sample data
The training sample data of indicating will, fingerpost as second level convolutional neural networks;
The training sample data are subjected to propagated forward and reverse conduction, convolution kernel and biasing are completed by gradient descent method
Update, alternately and repeatedly handle, until meeting the identification condition of convergence or until reach the frequency of training of requirement, training obtains feature
Vector;
By this feature vector training SVM classifier, 6 second level BaseNet models are obtained.
As shown in fig. 6, first order convolutional neural networks are rough sort, traffic sign is divided into 6 classes, such as: speed(-)limit sign, taboo
Enable mark, caution sign, round Warning Mark, rectangular Warning Mark, fingerpost;Second level convolutional neural networks are to previous stage
Rough sort result is classified again, and the corresponding output classification of 6 second level subdivision class models is respectively n1、n2、n3、n4、n5And n6, into
And realize the final Classification and Identification of n class Traffic Sign Images.
In addition, this method further include:
Using Kalman filtering and Camshift (Continuously Adaptive Mean-Shift, it is continuous adaptive
Answer MeanShift algorithm) algorithm tracks the Sign Board detected.Since determination is generally compared in the track of driving, use
Kalman filtering can effectively predicting tracing position, and Camshift algorithm being capable of the effectively distinctive colour of tracking mark board
Feature, and tracking velocity also meets the requirement of real-time.
The invention adopts the above technical scheme, the traffic sign recognition method in the rain and snow weather, comprising: to adopting
The image of collection is handled, the clear image after obtaining sleet;Sign board detection is carried out to the clear image;To inspection
The Sign Board image measured is identified board type identification.Technical solution proposed by the invention can be to the road in rain and snow weather
Road traffic sign is quickly accurately identified, and is advantageously accounted under bad weather, and driver can not since sight is hindered
The problem of timely and accurately capturing the flag information in road advantageously ensures that traffic safety and improves conevying efficiency.
It is understood that same or similar part can mutually refer in the various embodiments described above, in some embodiments
Unspecified content may refer to the same or similar content in other embodiments.
It should be noted that in the description of the present invention, term " first ", " second " etc. are used for description purposes only, without
It can be interpreted as indication or suggestion relative importance.In addition, in the description of the present invention, unless otherwise indicated, the meaning of " multiple "
Refer at least two.
Any process described otherwise above or method description are construed as in flow chart or herein, and expression includes
It is one or more for realizing specific logical function or process the step of executable instruction code module, segment or portion
Point, and the range of the preferred embodiment of the present invention includes other realization, wherein can not press shown or discussed suitable
Sequence, including according to related function by it is basic simultaneously in the way of or in the opposite order, Lai Zhihang function, this should be of the invention
Embodiment person of ordinary skill in the field understood.
It should be appreciated that each section of the invention can be realized with hardware, software, firmware or their combination.Above-mentioned
In embodiment, software that multiple steps or method can be executed in memory and by suitable instruction execution system with storage
Or firmware is realized.It, and in another embodiment, can be under well known in the art for example, if realized with hardware
Any one of column technology or their combination are realized: having a logic gates for realizing logic function to data-signal
Discrete logic, with suitable combinational logic gate circuit specific integrated circuit, programmable gate array (PGA), scene
Programmable gate array (FPGA) etc..
Those skilled in the art are understood that realize all or part of step that above-described embodiment method carries
It suddenly is that relevant hardware can be instructed to complete by program, the program can store in a kind of computer-readable storage medium
In matter, which when being executed, includes the steps that one or a combination set of embodiment of the method.
It, can also be in addition, each functional unit in each embodiment of the present invention can integrate in a processing module
It is that each unit physically exists alone, can also be integrated in two or more units in a module.Above-mentioned integrated mould
Block both can take the form of hardware realization, can also be realized in the form of software function module.The integrated module is such as
Fruit is realized and when sold or used as an independent product in the form of software function module, also can store in a computer
In read/write memory medium.
Storage medium mentioned above can be read-only memory, disk or CD etc..
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or spy described in conjunction with this embodiment or example
Point is included at least one embodiment or example of the invention.In the present specification, schematic expression of the above terms are not
Centainly refer to identical embodiment or example.Moreover, particular features, structures, materials, or characteristics described can be any
One or more embodiment or examples in can be combined in any suitable manner.
Although the embodiments of the present invention has been shown and described above, it is to be understood that above-described embodiment is example
Property, it is not considered as limiting the invention, those skilled in the art within the scope of the invention can be to above-mentioned
Embodiment is changed, modifies, replacement and variant.
Claims (14)
1. a kind of traffic sign recognition method in rain and snow weather characterized by comprising
The image of acquisition is handled, the clear image after obtaining sleet;
Sign board detection is carried out to the clear image;
Board type identification is identified to the Sign Board image detected.
2. the method according to claim 1, wherein the image of described pair of acquisition is handled, comprising:
It is coarse figure and detail view by the picture breakdown of acquisition;
The rainprint in detail view is removed, to obtain clear image.
3. according to the method described in claim 2, it is characterized in that, the picture breakdown by acquisition is coarse figure and details
Figure, comprising:
The image of acquisition is subjected to low-pass filtering treatment, the image obtained after processing is coarse figure;
The image of acquisition and coarse figure are subtracted each other, detail view is obtained.
4. according to the method described in claim 2, it is characterized in that, the sleet trace in the removal detail view includes:
Using the morphological difference between rainprint and traffic sign texture, the detail view is divided by sparse dictionary learning algorithm
For texture maps and rainprint figure.
5. according to the method described in claim 4, it is characterized in that, the sleet trace removed in detail view further include:
According to the prior information of rainprint shape feature, the texture maps and rainprint subgraph secondary sentence by rainprint length-width ratio
Not, come with more accurately decompositing texture maps from detail view.
6. the method according to claim 1, wherein it is described to the clear image carry out sign board detection,
Include:
Establish multilayer feature conspicuousness model;
The clear image is inputted into the multilayer feature conspicuousness model, to obtain sign board testing result.
7. according to the method described in claim 6, it is characterized in that, described establish multilayer feature conspicuousness model, comprising:
The distinctive information of sign board is extracted, is trained by the information input training pattern, and using boosting algorithm.
8. the method according to the description of claim 7 is characterized in that the distinctive information of the sign board includes:
Shape information, colouring information, gradient information and location information.
9. the method according to claim 1, wherein the described pair of Sign Board image detected is identified board class
Type identification, comprising:
Type identification is carried out to Sign Board using tandem type convolutional neural networks;
Wherein, the tandem type convolutional neural networks include: first order convolutional neural networks and second level convolutional neural networks;
The first order convolutional neural networks carry out rough sort to the Sign Board image of input, and the result of rough sort is input to
Second level convolutional neural networks are finely divided class, to identify the type of Sign Board.
10. according to the method described in claim 9, it is characterized in that, the tandem type convolutional neural networks carry out Sign Board
Type identification, detailed process include:
Establish training sample data;
Stable first order convolutional neural networks and second level convolutional neural networks are trained according to training sample data;
First order convolutional neural networks and second level convolutional neural networks are subjected to cascade and form tandem type convolutional neural networks;
Sign board type identification is carried out using the tandem type convolutional neural networks that training obtains.
11. according to the method described in claim 10, it is characterized in that, the training sample data include:
The Traffic Sign Images that actual acquisition arrives, and
Gaussian noise and the image after rotation and scaling processing are added to the Traffic Sign Images that actual acquisition arrives.
12. according to the method described in claim 10, it is characterized in that, described train stable according to training sample data
Level-one convolutional neural networks, comprising:
Training sample data are subjected to propagated forward and reverse conduction, convolution kernel and biasing are completed more by gradient descent method
Newly, alternately and repeatedly handle, until meeting the identification condition of convergence or until reach the frequency of training of requirement, training obtain feature to
Amount;
By this feature vector training SVM classifier, first order convolutional neural networks are obtained.
13. according to the method described in claim 10, it is characterized in that, described train stable according to training sample data
Second level convolutional neural networks, comprising:
Speed(-)limit sign, prohibitory sign, caution sign, round Warning Mark, rectangular indicateing arm are picked out in training sample data
The training sample data of will, fingerpost as second level convolutional neural networks;
The training sample data are subjected to propagated forward and reverse conduction, convolution kernel and biasing are completed more by gradient descent method
Newly, alternately and repeatedly handle, until meeting the identification condition of convergence or until reach the frequency of training of requirement, training obtain feature to
Amount;
By this feature vector training SVM classifier, second level convolutional neural networks are obtained.
14. according to claim 1 to 13 described in any item methods, which is characterized in that further include:
The Sign Board detected is tracked using Kalman filtering and Camshift algorithm.
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/CN2018/093095 WO2020000253A1 (en) | 2018-06-27 | 2018-06-27 | Traffic sign recognizing method in rain and snow |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110226170A true CN110226170A (en) | 2019-09-10 |
Family
ID=67822634
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201880005497.2A Pending CN110226170A (en) | 2018-06-27 | 2018-06-27 | A kind of traffic sign recognition method in rain and snow weather |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN110226170A (en) |
WO (1) | WO2020000253A1 (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111143553A (en) * | 2019-12-06 | 2020-05-12 | 国家计算机网络与信息安全管理中心 | Method and system for identifying specific information of real-time text data stream |
CN111178282A (en) * | 2019-12-31 | 2020-05-19 | 武汉中海庭数据技术有限公司 | Road traffic speed limit sign positioning and identifying method and device |
CN115661784A (en) * | 2022-10-12 | 2023-01-31 | 北京惠朗时代科技有限公司 | Intelligent traffic-oriented traffic sign image big data identification method and system |
Families Citing this family (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111260691B (en) * | 2020-01-18 | 2023-04-25 | 温州大学 | Space-time regular correlation filtering tracking method based on context awareness regression |
CN113076800A (en) * | 2021-03-03 | 2021-07-06 | 惠州市博实结科技有限公司 | Road sign board detection method and device |
CN113033687A (en) * | 2021-04-02 | 2021-06-25 | 西北工业大学 | Target detection and identification method under rain and snow weather condition |
CN113255555A (en) * | 2021-06-04 | 2021-08-13 | 清华大学 | Method, system, processing equipment and storage medium for identifying Chinese traffic sign board |
CN114495061B (en) * | 2022-01-25 | 2024-04-05 | 青岛海信网络科技股份有限公司 | Road traffic sign board identification method and device |
CN114818909B (en) * | 2022-04-22 | 2023-09-15 | 北大荒信息有限公司 | Weed detection method and device based on crop growth characteristics |
CN115830451B (en) * | 2022-12-02 | 2023-11-03 | 上海理工大学 | Smart city construction level identification method based on convolutional neural network |
CN115620265B (en) * | 2022-12-19 | 2023-04-18 | 华南理工大学 | Locomotive signboard information intelligent identification method and system based on deep learning |
CN117523535B (en) * | 2024-01-08 | 2024-04-12 | 浙江零跑科技股份有限公司 | Traffic sign recognition method, terminal equipment and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104850845A (en) * | 2015-05-30 | 2015-08-19 | 大连理工大学 | Traffic sign recognition method based on asymmetric convolution neural network |
CN105139344A (en) * | 2015-06-12 | 2015-12-09 | 中国科学院深圳先进技术研究院 | Method and system for removing influence of raindrops to single image based on frequency domain and phase consistency |
CN106919939A (en) * | 2017-03-14 | 2017-07-04 | 潍坊学院 | A kind of traffic signboard Tracking Recognition method and system |
CN106934374A (en) * | 2017-03-14 | 2017-07-07 | 潍坊学院 | The recognition methods of traffic signboard and system in a kind of haze scene |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3975138A1 (en) * | 2008-10-06 | 2022-03-30 | Mobileye Vision Technologies Ltd. | Bundling of driver assistance systems |
CN105844257B (en) * | 2016-04-11 | 2019-03-01 | 吉林大学 | The early warning system and method for road sign are missed based on machine vision travelling in fog day |
CN106056545A (en) * | 2016-05-24 | 2016-10-26 | 中国科学院深圳先进技术研究院 | Image rain removing method and image rain removing system |
CN106023112A (en) * | 2016-05-24 | 2016-10-12 | 中国科学院深圳先进技术研究院 | Image rain removing method and system based on wavelet analysis |
-
2018
- 2018-06-27 WO PCT/CN2018/093095 patent/WO2020000253A1/en active Application Filing
- 2018-06-27 CN CN201880005497.2A patent/CN110226170A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104850845A (en) * | 2015-05-30 | 2015-08-19 | 大连理工大学 | Traffic sign recognition method based on asymmetric convolution neural network |
CN105139344A (en) * | 2015-06-12 | 2015-12-09 | 中国科学院深圳先进技术研究院 | Method and system for removing influence of raindrops to single image based on frequency domain and phase consistency |
CN106919939A (en) * | 2017-03-14 | 2017-07-04 | 潍坊学院 | A kind of traffic signboard Tracking Recognition method and system |
CN106934374A (en) * | 2017-03-14 | 2017-07-07 | 潍坊学院 | The recognition methods of traffic signboard and system in a kind of haze scene |
Non-Patent Citations (1)
Title |
---|
WENCHENG WANG ET AL.: "Dehazing for images with large sky region", 《NEUROCOMPUTING》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111143553A (en) * | 2019-12-06 | 2020-05-12 | 国家计算机网络与信息安全管理中心 | Method and system for identifying specific information of real-time text data stream |
CN111143553B (en) * | 2019-12-06 | 2023-04-07 | 国家计算机网络与信息安全管理中心 | Method and system for identifying specific information of real-time text data stream |
CN111178282A (en) * | 2019-12-31 | 2020-05-19 | 武汉中海庭数据技术有限公司 | Road traffic speed limit sign positioning and identifying method and device |
CN115661784A (en) * | 2022-10-12 | 2023-01-31 | 北京惠朗时代科技有限公司 | Intelligent traffic-oriented traffic sign image big data identification method and system |
CN115661784B (en) * | 2022-10-12 | 2023-08-22 | 北京惠朗时代科技有限公司 | Intelligent traffic-oriented traffic sign image big data identification method and system |
Also Published As
Publication number | Publication date |
---|---|
WO2020000253A1 (en) | 2020-01-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110226170A (en) | A kind of traffic sign recognition method in rain and snow weather | |
CN105844257B (en) | The early warning system and method for road sign are missed based on machine vision travelling in fog day | |
Fang et al. | Road-sign detection and tracking | |
CN107133974B (en) | Gaussian Background models the vehicle type classification method combined with Recognition with Recurrent Neural Network | |
CN102968637B (en) | Complicated background image and character division method | |
CN108171112A (en) | Vehicle identification and tracking based on convolutional neural networks | |
Sheikh et al. | Traffic sign detection and classification using colour feature and neural network | |
Nandi et al. | Traffic sign detection based on color segmentation of obscure image candidates: a comprehensive study | |
CN109711264A (en) | A kind of bus zone road occupying detection method and device | |
Yang et al. | Improved lane detection with multilevel features in branch convolutional neural networks | |
CN110008900B (en) | Method for extracting candidate target from visible light remote sensing image from region to target | |
Sugiharto et al. | Traffic sign detection based on HOG and PHOG using binary SVM and k-NN | |
CN107704833A (en) | A kind of front vehicles detection and tracking based on machine learning | |
CN113837094A (en) | Road condition rapid analysis method based on full-color high-resolution remote sensing image | |
Jo | LED Dot matrix text recognition method in natural scene | |
Supreeth et al. | An approach towards efficient detection and recognition of traffic signs in videos using neural networks | |
CN108073940A (en) | A kind of method of 3D object instance object detections in unstructured moving grids | |
Joy et al. | Real time road lane detection using computer vision techniques in python | |
Varun et al. | A road traffic signal recognition system based on template matching employing tree classifier | |
CN106203368A (en) | A kind of traffic video frequency vehicle recognition methods based on SRC and SVM assembled classifier | |
Satti et al. | ICTS: Indian cautionary traffic sign classification using deep learning | |
Phu et al. | Traffic sign recognition system using feature points | |
Mohammed et al. | An overview on various methods of detection and recognition of traffic signs by Autonomous Vehicles | |
Shahbaz et al. | The Evaluation of Cascade Object Detector in Recognizing Different Samples of Road Signs | |
CN114299414B (en) | Vehicle red light running recognition and judgment method based on deep learning |
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: 20190910 |