CN109389167A - Traffic sign recognition method and system - Google Patents
Traffic sign recognition method and system Download PDFInfo
- Publication number
- CN109389167A CN109389167A CN201811151651.5A CN201811151651A CN109389167A CN 109389167 A CN109389167 A CN 109389167A CN 201811151651 A CN201811151651 A CN 201811151651A CN 109389167 A CN109389167 A CN 109389167A
- Authority
- CN
- China
- Prior art keywords
- image
- traffic sign
- pixel
- region
- original image
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
-
- 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
- G06V10/48—Extraction of image or video features by mapping characteristic values of the pattern into a parameter space, e.g. Hough transformation
-
- 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
- G06V10/56—Extraction of image or video features relating to colour
Abstract
A kind of traffic sign recognition method comprising following steps: S1, acquisition traffic sign original image extract the color characteristic and textural characteristics of traffic sign original image；S2, high-order spectrum signature and textural characteristics according to the traffic sign original image of extraction, carry out image-region coarse segmentation, traffic sign original image are divided into different pockets；The various features of the pocket image after segmentation are merged simultaneously；S3, similar image zonule is fused by new region according to image-region similarity measurement；S4, fused image-region is filtered, pass through filtering core setting area area threshold, interested target area is extracted, and detection fitting is carried out according to target area of the shape of image to extraction, the accurate circle extracted in image, triangle, rectangular board.
Description
Technical field
The present invention relates to technical field of transportation, in particular to a kind of traffic sign recognition method and system.
Background technique
Into several years, more researcher was engaged in the research of Traffic Sign Recognition both at home and abroad.But for complex scene
Traffic image, due to by illumination, block, rainy weather etc. influences, usually will cause the higher omission factor of traffic sign and mistake
Inspection rate.Common traffic sign recognition method is divided into both direction, is identified based on color threshold method, but due to color
Each channel correlation is easy to be influenced by illumination etc., causes the imperfection of Traffic Sign Recognition.Traffic sign based on shape
Recognizer has certain robustness to illumination effect, and traffic sign is divided into several regions and carries out edge detection, then sharp
Carry out Traffic Sign Recognition with shape feature, achieve ideal effect, but due to being deformed under complex scene, day
Gas is blocked etc. and to be influenced, and will also result in the not accuracy of Traffic Sign Recognition.In recent years, quickly growing with deep learning, base
It is more and more studied in the traffic sign recognition method of deep learning, obtains extraordinary detection effect.But based on nerve
The method for traffic sign detection of network needs a large amount of labeled data, and cost consumption is larger.
Summary of the invention
In view of this, the present invention proposes a kind of traffic sign recognition method and system.
A kind of traffic sign recognition method comprising following steps:
S1, acquisition traffic sign original image, extract the color characteristic and textural characteristics of traffic sign original image；
S2, high-order spectrum signature and textural characteristics according to the traffic sign original image of extraction, carry out image-region rough segmentation
It cuts, traffic sign original image is divided into different pockets；Simultaneously to a variety of spies of the pocket image after segmentation
Sign is merged；
S3, similar image zonule is fused by new region according to image-region similarity measurement；
S4, fused image-region is filtered, by filtering core setting area area threshold, extracts interested target
Region, and detection fitting is carried out according to target area of the shape of image to extraction, the accurate circle extracted in image, triangle
Shape, rectangular board.
In traffic sign recognition method of the present invention,
The step S1 includes:
It is converted by Gabor wavelet and carries out texture feature extraction.
In traffic sign recognition method of the present invention,
Gabor function g (x, y) is indicated are as follows:
The Fourier transform G (u, v) of Gabor function g (x, y) is as follows:
Wherein,
Given original image I (x, y), Gabor wavelet convert Wm,n(u, v) is as follows:
Wherein subscript * indicates complex conjugate, and subscript m and n respectively indicate multiple scales and direction；σ indicates standard deviation, σx、σy、
σm、σyRespectively indicate the variance of coordinate x, y and scale m, direction n；
Then the mean μ on the multiple scales of image and direction is calculatedm,nAnd standard deviation sigmam,n；
μm,n=∫ ∫ | Wm,n(x,y)|dxdy
Final Gabor textural characteristics are expressed as T=[u0,0σ0,0u0,1…uM-1,N-1σM-1,N-1]。
In traffic sign recognition method of the present invention,
The step S1 includes:
The color characteristic for extracting traffic sign original image indicates are as follows: C=[RMean,RVar,GMean,GVar,BMean,BVar]T；
Wherein, wherein Var indicate neighborhood of pixel points class variance, RVar、GVar、BVarThe channel R of expression pixel p neighborhood,
The variance in the channel G, the channel V；Wherein Mean indicates the mean value of neighborhood of pixel points class, RMean、GMean、BMeanIndicate that pixel p is adjacent
The mean value in the channel R in domain, the channel G, the channel V.
In traffic sign recognition method of the present invention,
Pre-segmentation is carried out to image by the method for image clustering Meanshift in the step S2, comprising:
For image slices vegetarian refreshments p, using pixel p as the center of circle, h is radius, falls in the pixel p of round classiDefine two rule
Then: closer pixel, probability density are higher with pixel p color；Closer pixel, probability with the position pixel p
Density is higher；Then two regular probability density are multiplied as follows:
Wherein K represents probability density, and r represents color, behalf position；Indicate pixel p and pixel piPosition
Difference is set,Indicate pixel p and pixel piColour-difference；For Position probability densities, pixel p and
Pixel piPosition is closer, and Position probability densities are bigger；For color probability density, pixel p and pixel pi
Color is closer, and color probability density is bigger；
Then according to probability density productPixel in image is clustered, as Meanshift cluster
Method.
In traffic sign recognition method of the present invention,
The various features of pocket image after segmentation are merged: F=[ω1C+(1-ω1) T], ω1For weight
Coefficient, C are color characteristic, and T is Gabor textural characteristics, and F is fused feature.
In traffic sign recognition method of the present invention,
Image subblock zonule r after image pre-segmentation divides in the step S3iWith sub-block zonule rjColor and line
Manage similarity measurement is defined as:
Wherein | | Fm-Fn| | it is measured using traditional Euclidean distance, m is image subblock zonule riIn pixel, n
For image subblock zonule rjIn pixel；M and N is respectively image subblock zonule riAnd rjThe total amount of middle pixel.
In traffic sign recognition method of the present invention,
Lead to Hough transformation in the step S4 and carries out SHAPE DETECTION fitting；
Its process is to obtain the marginal information in targets of interest region, the straight line and circle at target area edge is examined, according to rectangle
The angle of the four edges of board, the angle information on Atria side screen the straight line detected, so that it is determined that rectangle out
Traffic mark board, triangle traffic sign board and circular traffic sign board；
Straight line y=m in image space0x+b0A point in the corresponding space Hough, the pixel of image space
(x0,y0) it is mapped as a line in the space Hough, i.e. b=-x0m+y0；Pixel (x is crossed in the space Hough0,y0), (x1,y1)
Straight line indicated with the intersection point of two lines in the space Hough；Vertical line is indicated with polar coordinates x=rcos θ+rsin θ.It passes through in this way
Hough transform is crossed, each pixel (x, y) in image is mapped as the sine curve in space (r, θ) of Hough, and image is empty
Between in Hough corresponding to conllinear point the space (r, θ) in sine curve intersect at point (r', θ ')；By solving (r, θ)
Most curves is to carry out straight-line detection；Round equation is become by (x-a) using Hough transformation detection bowlder2+(y-b)2=r2
It is solved.
The present invention also provides a kind of Traffic Sign Recognition Systems comprising such as lower unit:
Feature extraction unit extracts the color characteristic of traffic sign original image for acquiring traffic sign original image
And textural characteristics；
Image-region coarse extraction unit, for special according to the high-order spectrum signature and texture of the traffic sign original image of extraction
Sign carries out image-region coarse segmentation, traffic sign original image is divided into different pockets；Simultaneously to small after segmentation
The various features of block area image are merged；
Image Region Merging unit, for being fused into similar image zonule newly according to image-region similarity measurement
Region；
The accurate extraction unit of traffic mark board passes through filtering core setting area for filtering fused image-region
Area threshold extracts interested target area, and carries out detection fitting according to target area of the shape of image to extraction, essence
Circle, triangle, rectangular board in true extraction image.
Implement traffic sign recognition method provided by the invention and system has the advantages that compared with prior art
Traffic Sign Recognition is carried out for the Traffic Sign Images of complex scene, it at a lower cost can quickly, efficiently and accurately
Circle, rectangle, triangle traffic sign board in identification image.
Detailed description of the invention
Fig. 1 is traffic sign recognition method flow chart；
Fig. 2 is the experimental image final result image that Traffic Sign Recognition pilot process generates.
Specific embodiment are as follows:
A kind of traffic sign recognition method comprising following steps:
S1, acquisition traffic sign original image, extract the color characteristic and textural characteristics of traffic sign original image；
S2, high-order spectrum signature and textural characteristics according to the traffic sign original image of extraction, carry out image-region rough segmentation
It cuts, traffic sign original image is divided into different pockets；Simultaneously to a variety of spies of the pocket image after segmentation
Sign is merged；
Image coarse segmentation is carried out by textural characteristics, can effectively be clustered similar element to the same region, drop
The difficulty of low subsequent calculating.Meanwhile it being handled by the way that Traffic Sign Images are divided into certain zonule instead of element, energy
Enough accelerate calculating speed.
The textural characteristics and high-order spectrum signature of blending image, textural characteristics and high-order spectrum signature significantly more efficient can be handled
The inapparent region of background/target in target image, the shadow of picture blur caused by capable of effectively excluding because of illumination, raindrop etc.
It rings.
S3, similar image zonule is fused by new region according to image-region similarity measurement；
The zonule divided in Traffic Sign Images is merged according to the feature of fusion, and to fused region
Filtering, can effectively extract interested target area in image.
S4, fused image-region is filtered, by filtering core setting area area threshold, extracts interested target
Region, and detection fitting is carried out according to target area of the shape of image to extraction, the accurate circle extracted in image, triangle
Shape, rectangular board.
By the form fit of the area-of-interest to extraction, checked using Hough transformation etc. circle in fitted area,
Triangle, rectangle can accurately extract circle in target image, rectangle, triangle sign board.
In traffic sign recognition method of the present invention,
The step S1 includes:
It is converted by Gabor wavelet and carries out texture feature extraction.
In traffic sign recognition method of the present invention,
Gabor function g (x, y) is indicated are as follows:
The Fourier transform G (u, v) of Gabor function g (x, y) is as follows:
Wherein,
Given original image I (x, y), Gabor wavelet convert Wm,n(u, v) is as follows:
Wherein subscript * indicates complex conjugate, and subscript m and n respectively indicate multiple scales and direction；σ indicates standard deviation, σx、σy、
σm、σyRespectively indicate the variance of coordinate x, y and scale m, direction n；
Then the mean μ on the multiple scales of image and direction is calculatedm,nAnd standard deviation sigmam,n；
μm,n=∫ ∫ | Wm,n(x,y)|dxdy
Final Gabor textural characteristics are expressed as T=[u0,0σ0,0u0,1…uM-1,N-1σM-1,N-1]。
In traffic sign recognition method of the present invention,
The step S1 includes:
The color characteristic for extracting traffic sign original image indicates are as follows: C=[RMean,RVar,GMean,GVar,BMean,BVar]T；
Wherein, wherein Var indicate neighborhood of pixel points class variance, RVar、GVar、BVarThe channel R of expression pixel p neighborhood,
The variance in the channel G, the channel V；Wherein Mean indicates the mean value of neighborhood of pixel points class, RMean、GMean、BMeanIndicate that pixel p is adjacent
The mean value in the channel R in domain, the channel G, the channel V.
In traffic sign recognition method of the present invention,
Pre-segmentation is carried out to image by the method for image clustering Meanshift in the step S2, comprising:
For image slices vegetarian refreshments p, using pixel p as the center of circle, h is radius, falls in the pixel p of round classiDefine two rule
Then: closer pixel, probability density are higher with pixel p color；Closer pixel, probability with the position pixel p
Density is higher；Then two regular probability density are multiplied as follows:
Wherein K represents probability density, and r represents color, behalf position；Indicate pixel p and pixel piPosition
Difference is set,Indicate pixel p and pixel piColour-difference；For Position probability densities, pixel p and
Pixel piPosition is closer, and Position probability densities are bigger；For color probability density, pixel p and pixel pi
Color is closer, and color probability density is bigger；
Then according to probability density productPixel in image is clustered, as Meanshift cluster
Method.
In traffic sign recognition method of the present invention,
The various features of pocket image after segmentation are merged: F=[ω1C+(1-ω1) T], ω1For weight
Coefficient, C are color characteristic, and T is Gabor textural characteristics, and F is fused feature.
In traffic sign recognition method of the present invention,
Image subblock zonule r after image pre-segmentation divides in the step S3iWith sub-block zonule rjColor and line
Manage similarity measurement is defined as:
Wherein | | Fm-Fn| | it is measured using traditional Euclidean distance, m is image subblock zonule riIn pixel, n
For image subblock zonule rjIn pixel；M and N is respectively image subblock zonule riAnd rjThe total amount of middle pixel.
In traffic sign recognition method of the present invention,
Lead to Hough transformation in the step S4 and carries out SHAPE DETECTION fitting；
Its process is to obtain the marginal information in targets of interest region, the straight line and circle at target area edge is examined, according to rectangle
The angle of the four edges of board, the angle information on Atria side screen the straight line detected, so that it is determined that rectangle out
Traffic mark board, triangle traffic sign board and circular traffic sign board；
Straight line y=m in image space0x+b0A point in the corresponding space Hough, the pixel of image space
(x0,y0) it is mapped as a line in the space Hough, i.e. b=-x0m+y0；Pixel (x is crossed in the space Hough0,y0), (x1,y1)
Straight line indicated with the intersection point of two lines in the space Hough；Vertical line is indicated with polar coordinates x=rcos θ+rsin θ.It passes through in this way
Hough transform is crossed, each pixel (x, y) in image is mapped as the sine curve in space (r, θ) of Hough, and image is empty
Between in Hough corresponding to conllinear point the space (r, θ) in sine curve intersect at point (r', θ ')；By solving (r, θ)
Most curves is to carry out straight-line detection；Round equation is become by (x-a) using Hough transformation detection bowlder2+(y-b)2=r2
It is solved.
The present invention also provides a kind of Traffic Sign Recognition Systems comprising such as lower unit:
Feature extraction unit extracts the color characteristic of traffic sign original image for acquiring traffic sign original image
And textural characteristics；
Image-region coarse extraction unit, for special according to the high-order spectrum signature and texture of the traffic sign original image of extraction
Sign carries out image-region coarse segmentation, traffic sign original image is divided into different pockets；Simultaneously to small after segmentation
The various features of block area image are merged；
Image Region Merging unit, for being fused into similar image zonule newly according to image-region similarity measurement
Region；
The accurate extraction unit of traffic mark board passes through filtering core setting area for filtering fused image-region
Area threshold extracts interested target area, and carries out detection fitting according to target area of the shape of image to extraction, essence
Circle, triangle, rectangular board in true extraction image.
As shown in Figure 1, 2, obtain the original traffic sign image that need to handle first, extract the image color characteristic and
Gabor textural characteristics；Secondly original traffic image is divided by image progress coarse segmentation using Meanshift a certain number of
Subregion, color characteristic and textural characteristics in combination with original image carry out Fusion Features；Again by the figure after segmentation
As subregion progress similarity measurement, similar region is fused into bigger target area；Finally to the target area of extraction
It is screened, extracts interesting target region, and accurately extract in area-of-interest by the fitting of Hough transformation SHAPE DETECTION
Triangle, rectangle, circular sign board.Its traffic sign recognition method overall flow figure is as shown in Figure 1.
Fig. 2 is the experimental image final result image that Traffic Sign Recognition pilot process generates.
The primary picture generated during Traffic Sign Recognition System has:
First is classified as the Traffic Sign Images being originally inputted in Fig. 2；Second is classified as through Meanshift pre-segmentation in Fig. 2
It is divided into the image of certain amount subregion afterwards；Third is classified as laggard by region similarity measurement progress region fusion in Fig. 2
The interesting target administrative division map that row simple screening extracts；The 4th is classified as by extracting most after Hough transformation detection fitting in Fig. 2
Whole triangle, rectangle, circular traffic sign result images.
Table 1 is the discrimination testing 100 images and obtaining.
It is understood that for those of ordinary skill in the art, can do in accordance with the technical idea of the present invention
Various other changes and modifications out, and all these changes and deformation all should belong to the protection model of the claims in the present invention
It encloses.
Claims (9)
1. a kind of traffic sign recognition method, which is characterized in that it includes the following steps:
S1, acquisition traffic sign original image, extract the color characteristic and textural characteristics of traffic sign original image；
S2, high-order spectrum signature and textural characteristics according to the traffic sign original image of extraction, carry out image-region coarse segmentation, will
Traffic sign original image is divided into different pockets；The various features of the pocket image after segmentation are carried out simultaneously
Fusion；
S3, similar image zonule is fused by new region according to image-region similarity measurement；
S4, fused image-region is filtered, by filtering core setting area area threshold, extracts interested target area
Domain, and detection fitting is carried out according to target area of the shape of image to extraction, the accurate circle extracted in image, triangle
Shape, rectangular board.
2. traffic sign recognition method as described in claim 1, which is characterized in that
The step S1 includes:
It is converted by Gabor wavelet and carries out texture feature extraction.
3. traffic sign recognition method as claimed in claim 2, which is characterized in that
Gabor function g (x, y) is indicated are as follows:
The Fourier transform G (u, v) of Gabor function g (x, y) is as follows:
Wherein,
Given original image I (x, y), Gabor wavelet convert Wm,n(u, v) is as follows:
Wherein subscript * indicates complex conjugate, and subscript m and n respectively indicate multiple scales and direction；σ indicates standard deviation, σx、σy、σm、σy
Respectively indicate the variance of coordinate x, y and scale m, direction n；
Then the mean μ on the multiple scales of image and direction is calculatedm,nAnd standard deviation sigmam,n；
μm,n=∫ ∫ | Wm,n(x,y)|dxdy
Final Gabor textural characteristics are expressed as T=[u0,0σ0,0u0,1…uM-1,N-1σM-1,N-1]。
4. traffic sign recognition method as claimed in claim 3, which is characterized in that
The step S1 includes:
The color characteristic for extracting traffic sign original image indicates are as follows: C=[RMean,RVar,GMean,GVar,BMean,BVar]T；
Wherein, wherein Var indicate neighborhood of pixel points class variance, RVar、GVar、BVarIndicate that the channel R, the G of pixel p neighborhood are logical
The variance in road, the channel V；Wherein Mean indicates the mean value of neighborhood of pixel points class, RMean、GMean、BMeanIndicate pixel p neighborhood
The channel R, the channel G, the channel V mean value.
5. traffic sign recognition method as claimed in claim 4, which is characterized in that
Pre-segmentation is carried out to image by the method for image clustering Meanshift in the step S2, comprising:
For image slices vegetarian refreshments p, using pixel p as the center of circle, h is radius, falls in the pixel p of round classiDefine two rules: with
The closer pixel of pixel p color, probability density are higher；Closer pixel, probability density are got over the position pixel p
It is high；Then two regular probability density are multiplied as follows:
Wherein K represents probability density, and r represents color, behalf position；Indicate pixel p and pixel piAlternate position spike,Indicate pixel p and pixel piColour-difference；For Position probability densities, pixel p and pixel
piPosition is closer, and Position probability densities are bigger；For color probability density, pixel p and pixel piColor is got over
Closely, color probability density is bigger；
Then according to probability density productPixel in image is clustered, as Meanshift clustering method.
6. traffic sign recognition method as claimed in claim 5, which is characterized in that
The various features of pocket image after segmentation are merged: F=[ω1C+(1-ω1) T], ω1For weight coefficient,
C is color characteristic, and T is Gabor textural characteristics, and F is fused feature.
7. traffic sign recognition method as claimed in claim 6, which is characterized in that
Image subblock zonule r after image pre-segmentation divides in the step S3iWith sub-block zonule rjColor and texture phase
Like property measure definitions are as follows:
Wherein | | Fm-Fn| | it is measured using traditional Euclidean distance, m is image subblock zonule riIn pixel, n be figure
As sub-block zonule rjIn pixel；M and N is respectively image subblock zonule riAnd rjThe total amount of middle pixel.
8. traffic sign recognition method as claimed in claim 7, which is characterized in that
Lead to Hough transformation in the step S4 and carries out SHAPE DETECTION fitting；
Its process is to obtain the marginal information in targets of interest region, the straight line and circle at target area edge is examined, according to rectangle board
The angle of four edges, the angle information on Atria side screen the straight line detected, so that it is determined that rectangle traffic out
Sign board, triangle traffic sign board and circular traffic sign board；
Straight line y=m in image space0x+b0A point in the corresponding space Hough, the pixel (x of image space0,
y0) it is mapped as a line in the space Hough, i.e. b=-x0m+y0；Pixel (x is crossed in the space Hough0,y0), (x1,y1)
Straight line is indicated with the intersection point of two lines in the space Hough；Vertical line is indicated with polar coordinates x=rcos θ+rsin θ.Pass through in this way
Hough transform, each pixel (x, y) in image are mapped as the sine curve in space (r, θ) of Hough, image space
In Hough corresponding to conllinear point the space (r, θ) in sine curve intersect at point (r', θ ')；By solving (r, θ) most
More curves is to carry out straight-line detection；Round equation is become by (x-a) using Hough transformation detection bowlder2+(y-b)2=r2Into
Row solves.
9. a kind of Traffic Sign Recognition System, which is characterized in that it includes such as lower unit:
Feature extraction unit extracts the color characteristic and line of traffic sign original image for acquiring traffic sign original image
Manage feature；
Image-region coarse extraction unit, for the high-order spectrum signature and textural characteristics according to the traffic sign original image of extraction,
Image-region coarse segmentation is carried out, traffic sign original image is divided into different pockets；Simultaneously to the fritter after segmentation
The various features of area image are merged；
Image Region Merging unit, for similar image zonule to be fused into new area according to image-region similarity measurement
Domain；
The accurate extraction unit of traffic mark board passes through filtering core setting area area for filtering fused image-region
Threshold value extracts interested target area, and carries out detection fitting according to target area of the shape of image to extraction, accurately
Circle, triangle, rectangular board in extraction image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811151651.5A CN109389167A (en) | 2018-09-29 | 2018-09-29 | Traffic sign recognition method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811151651.5A CN109389167A (en) | 2018-09-29 | 2018-09-29 | Traffic sign recognition method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109389167A true CN109389167A (en) | 2019-02-26 |
Family
ID=65419043
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811151651.5A Pending CN109389167A (en) | 2018-09-29 | 2018-09-29 | Traffic sign recognition method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109389167A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110399771A (en) * | 2019-04-12 | 2019-11-01 | 泰州阿法光电科技有限公司 | Traffic environment on-site maintenance system |
| CN111666811A (en) * | 2020-04-22 | 2020-09-15 | 北京联合大学 | Method and system for extracting traffic sign area in traffic scene image |
| CN112464731A (en) * | 2020-11-03 | 2021-03-09 | 南京理工大学 | Traffic sign detection and identification method based on image processing |
| CN113361303A (en) * | 2020-03-05 | 2021-09-07 | 百度在线网络技术（北京）有限公司 | Temporary traffic sign board identification method, device and equipment |
| CN115035004A (en) * | 2022-04-15 | 2022-09-09 | 腾讯科技（深圳）有限公司 | Image processing method, apparatus, device, readable storage medium and program product |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106296680A (en) * | 2016-08-08 | 2017-01-04 | 长安大学 | A kind of multiple features fusion high-resolution remote sensing image dividing method based on region |
| US20170193313A1 (en) * | 2016-01-04 | 2017-07-06 | Texas Instruments Incorporated | Real time traffic sign recognition |
| CN107577981A (en) * | 2016-07-04 | 2018-01-12 | 高德信息技术有限公司 | A kind of road traffic index identification method and device |
| CN107909059A (en) * | 2017-11-30 | 2018-04-13 | 中南大学 | It is a kind of towards cooperateing with complicated City scenarios the traffic mark board of bionical vision to detect and recognition methods |
-
2018
- 2018-09-29 CN CN201811151651.5A patent/CN109389167A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170193313A1 (en) * | 2016-01-04 | 2017-07-06 | Texas Instruments Incorporated | Real time traffic sign recognition |
| CN107577981A (en) * | 2016-07-04 | 2018-01-12 | 高德信息技术有限公司 | A kind of road traffic index identification method and device |
| CN106296680A (en) * | 2016-08-08 | 2017-01-04 | 长安大学 | A kind of multiple features fusion high-resolution remote sensing image dividing method based on region |
| CN107909059A (en) * | 2017-11-30 | 2018-04-13 | 中南大学 | It is a kind of towards cooperateing with complicated City scenarios the traffic mark board of bionical vision to detect and recognition methods |
Non-Patent Citations (7)
| Title |
|---|
| ZHI-CHUN HUANG 等: "CONTENT-BASED IMAGE RETRIEVAL USING COLOR MOMENT AND GABOR TEXTURE FEATURE", 《IEEE》 * |
| 张建华 等: "基于Gabor小波和颜色矩的棉花盲椿象分类方法", 《农业工程学报》 * |
| 杨帆 等: "《数字图像处理与分析（第3版）》", 31 May 2015 * |
| 王更 等: "融合颜色－纹理模型的均值漂移分割算法", 《测绘科学》 * |
| 秦恩泉: "基于显著图的交通标志检测方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 蔡成涛 等: "《海洋环境下的计算机视觉技术》", 31 October 2015 * |
| 邓超 等: "《数字图像处理与模式识别研究》", 30 June 2018 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110399771A (en) * | 2019-04-12 | 2019-11-01 | 泰州阿法光电科技有限公司 | Traffic environment on-site maintenance system |
| CN110399771B (en) * | 2019-04-12 | 2020-09-11 | 厦门瞳景智能科技有限公司 | Traffic environment field maintenance system |
| CN113361303A (en) * | 2020-03-05 | 2021-09-07 | 百度在线网络技术（北京）有限公司 | Temporary traffic sign board identification method, device and equipment |
| CN113361303B (en) * | 2020-03-05 | 2023-06-23 | 百度在线网络技术（北京）有限公司 | Temporary traffic sign board identification method, device and equipment |
| CN111666811A (en) * | 2020-04-22 | 2020-09-15 | 北京联合大学 | Method and system for extracting traffic sign area in traffic scene image |
| CN111666811B (en) * | 2020-04-22 | 2023-08-15 | 北京联合大学 | Method and system for extracting traffic sign board area in traffic scene image |
| CN112464731A (en) * | 2020-11-03 | 2021-03-09 | 南京理工大学 | Traffic sign detection and identification method based on image processing |
| CN115035004A (en) * | 2022-04-15 | 2022-09-09 | 腾讯科技（深圳）有限公司 | Image processing method, apparatus, device, readable storage medium and program product |
| CN115035004B (en) * | 2022-04-15 | 2023-02-10 | 腾讯科技（深圳）有限公司 | Image processing method, apparatus, device, readable storage medium and program product |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109389167A (en) | Traffic sign recognition method and system | |
| CN107610114B (en) | optical satellite remote sensing image cloud and snow fog detection method based on support vector machine | |
| CN105844228B (en) | A kind of remote sensing images cloud detection method of optic based on convolutional neural networks | |
| CN101599120B (en) | Identification method of remote sensing image building | |
| Tao et al. | Unsupervised detection of built-up areas from multiple high-resolution remote sensing images | |
| Liu et al. | Automated vehicle extraction and speed determination from QuickBird satellite images | |
| CN109614936B (en) | Layered identification method for remote sensing image airplane target | |
| CN102254159B (en) | Interpretation method for digital readout instrument | |
| CN107392141A (en) | A kind of airport extracting method based on conspicuousness detection and LSD straight-line detections | |
| CN109635733B (en) | Parking lot and vehicle target detection method based on visual saliency and queue correction | |
| CN108052904B (en) | Method and device for acquiring lane line | |
| CN103530600A (en) | License plate recognition method and system under complicated illumination | |
| CN104573685A (en) | Natural scene text detecting method based on extraction of linear structures | |
| CN107688777B (en) | Urban green land extraction method for collaborative multi-source remote sensing image | |
| CN103473551A (en) | Station logo recognition method and system based on SIFT operators | |
| CN103984946A (en) | High resolution remote sensing map road extraction method based on K-means | |
| CN106897681A (en) | A kind of remote sensing images comparative analysis method and system | |
| CN106339707A (en) | Instrument pointer image recognition method based on symmetrical characteristics | |
| An et al. | An automated airplane detection system for large panchromatic image with high spatial resolution | |
| CN108629286A (en) | A kind of remote sensing airport target detection method based on the notable model of subjective perception | |
| CN106845542A (en) | Paper money number intelligent identification Method based on DSP | |
| CN108509950B (en) | Railway contact net support number plate detection and identification method based on probability feature weighted fusion | |
| CN107480585A (en) | Object detection method based on DPM algorithms | |
| CN110738216A (en) | Medicine identification method based on improved SURF algorithm | |
| CN109635722B (en) | Automatic identification method for high-resolution remote sensing image intersection |
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 |
Application publication date: 20190226 |
|
| RJ01 | Rejection of invention patent application after publication |