CN107301405A - Method for traffic sign detection under natural scene - Google Patents
Method for traffic sign detection under natural scene Download PDFInfo
- Publication number
- CN107301405A CN107301405A CN201710540228.3A CN201710540228A CN107301405A CN 107301405 A CN107301405 A CN 107301405A CN 201710540228 A CN201710540228 A CN 201710540228A CN 107301405 A CN107301405 A CN 107301405A
- Authority
- CN
- China
- Prior art keywords
- scene
- image
- region
- traffic sign
- color
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention proposes the method for traffic sign detection under a kind of natural scene, including：Obtain the detection image shot under natural scene；Monochrome information to detection image is counted, different luminance areas are marked off according to grade brightness threshold value, the pixel ratio of different luminance areas is calculated respectively, and image is divided into by dark scene, bright scene, backlight scene and normal scene according to each pixel ratio and scene classification threshold value；Gamma parameter values are chosen according to scene classification result, strengthen algorithm using adaptive Gamma carries out image enhancement processing to classification chart picture；Under RGB color, choose partitioning algorithm for different scenes and carry out segmentation of a color image, obtain suspected target region；Gray level image after color segmentation is subjected to binary conversion treatment, the suspected target region after binaryzation is obtained；Suspected target region is screened by Feature Selection device, positioning traffic sign region.The robustness and real-time of Mark Detection can be taken into account.
Description
Technical field
The present invention relates to a kind of inspection of the traffic sign under landmark identification detection technique field, more particularly to natural scene
Survey method.
Background technology
Traffic sign plays an important roll for ensureing that road traffic is current in order, in driving procedure, accurately identifies friendship
Logical mark is significant to lifting road drive safety, and traffic accident may be caused by ignoring traffic sign.For nobody
For driving, if vehicle is unable to automatic detection and traffic sign is recognized accurately, nothing truly can not be turned into
People drives.At present, in computer vision, pattern-recognition, the field such as intelligent robot and intelligent transportation system, road traffic sign detection
Widely studied with identification technology, with important academic significance and practical value.Road traffic sign detection is recognized in reality
Applied in life and still face huge challenge, including：ADAS, the many-side such as automatic Pilot and mark monitoring maintenance.Cause traffic mark
The difficult factor of will detection identification has many, including：(1) sign board is chronically exposed under sunlight and causes discoloration；(2) it is empty
Gas pollutes and weather conditions cause visibility to reduce；(3) outdoor complicated light conditions influence sign board color；(4) sign board quilt
Barrier is blocked；(5) vehicle operation causes motion blur.Interference all brings difficulty to Mark Detection identification above, to correlation calculation
The real-time and robustness of method propose higher requirement.
At present, road traffic sign detection module main approaches include being based on color segmentation technique, based on given shape
Detection technique, and the method such as the grader based on texture and local characteristic Design.
Include extracting target area under RGB, HSI and YUV color spaces respectively based on color segmentation technique.(1) it is based on
Color segmentation under rgb space, by using R, G, the difference relationship between tri- components of B extracts color of object region.However,
RGB color is very sensitive to illumination variation, easily causes missing inspection or false retrieval situation, therefore how to overcome this problem to be always
The emphasis and difficult point of color segmentation under rgb space.(2) based on the color segmentation under HSI spaces, RGB color is converted into
HSI color spaces, color of object region is extracted the characteristics of using HSI color spaces to illumination-insensitive.However, color space
Conversion is a time-consuming process, and the real-time that cause detection process cannot be ensured.(3) based on the color under yuv space point
Cut, RGB color is converted into YUV color spaces, the system will be used to detect blue rectangle sign board.However, for planting
The various sign board of class, it is not enough that this method highlights robustness.Segmentation based on color is led to needs design often in pursuit robustness
Many threshold values, and the determination of these threshold values is being extremely difficult in practice.Had based on color segmentation under rgb space and meet real
The advantage of when property requirement, but how to meet emphasis and be worth the place of breakthrough that robustness requirement is always the area research.
Detection technique based on given shape is to utilize the specific shape facility of sign board, target area is extracted, based on shape
The algorithm of shape feature will be applied directly on image, mainly be included：Hough transform, radial symmetric and histogram of gradients etc..
(1) Hough transform is mainly used to detect circle marker, is not enough to detect all character shapes.In addition, Hough transform is also present
Calculate slower shortcoming, it is impossible to meet requirement of real-time.(2) radial symmetric fast algorithm of detecting utilizes triangle, square, water chestnut
Shape, circular symmetry extracts target area, however, for the target of deformation, exposing robustness not enough.(3) gradient Nogata
Figure make use of target signature to calculate the characteristics of all relying on image gradient information, carry out Objective extraction.However, Gradient Features are to figure
As noise is very sensitive, it is impossible to meet robustness requirement.Detection technique based on given shape is generally extracting shape facility mistake
Larger time cost is expended in journey, it is impossible to meet requirement of real-time.
Grader based on texture and local characteristic Design is to be complementary to one another color characteristic with shape facility, first with
Color segmentation carries out rough detection, recycles shape facility to carry out smart detection, mainly includes：HSI color segmentations and template matches knot
Close；Color-shape pair；Neutral net；Adaboost graders；HOG and SVM classifier are combined.(1) HSI color segmentations and template
Matching process, completes segmentation first in HSI color spaces, recycles shape template matching process to obtain testing result.However,
HSI colors are changed and template matches will all take considerable time, it is impossible to meet requirement of real-time.(2) color-shape pair is to be directed to
Different traffic signs connects specific color segmentation result with specific shape extracting method.But for urban transportation
Complex jamming, robustness it cannot be guaranteed that.(3) neutral net is to separately design a neutral net spy for color and shape feature
Levying extractor recycles fuzzy logic to merge it, but this method amount of calculation is very big, it is impossible to meet requirement of real-time.(4)
Adaboost grader efficient combination colors and local characteristic information realize road traffic sign detection, but exist for complicated reality scene
Line computation needs the long period, it is impossible to meet requirement of real-time.(5) it is each color component of component that HOG and SVM classifier, which are combined,
HOG features, colouring information and edge information fusion are effectively risen to the input feature vector for being used as SVM classifier, but the process
It is time-consuming longer, it is impossible to meet requirement of real-time.
The studies above method can not all take into account the robustness and real-time of Mark Detection, often ensureing robustness, improving
On the premise of Detection accuracy, it is impossible to meet requirement of real-time；Also or under the premise of real-time is ensured, for actual environment Shandong
Rod sex expression is not enough.
The content of the invention
The technical problems to be solved by the invention are to provide the method for traffic sign detection under a kind of natural scene, can take into account
The robustness and real-time of Mark Detection.
To solve the above problems, the present invention proposes the method for traffic sign detection under a kind of natural scene, including following step
Suddenly：
S1：Obtain the detection image shot under natural scene；
S2：Monochrome information to the detection image is counted, and different brightness regions are marked off according to grade brightness threshold value
Domain, calculates the pixel ratio of different luminance areas respectively, and image is divided into the moon according to each pixel ratio and scene classification threshold value
Dark scene, bright scene, backlight scene and normal scene；
S3：Gamma parameter values are chosen according to scene classification result, algorithm is strengthened to classification chart using adaptive Gamma
As carrying out image enhancement processing；
S4：Under RGB color, choose partitioning algorithm for different scenes and carry out segmentation of a color image, obtain doubtful
Target area；Wherein, for bright scene, color segmentation is carried out using normalization RGB partitioning algorithms；For dark scene, inverse
Light field scape and normal scene, carry out color segmentation, the improved three-component chromatism method using improved three-component chromatism method
By setting adaptive weighted factor come the adaptive color region for adjusting extraction；
S5：Gray level image after color segmentation is subjected to binary conversion treatment, the suspected target region after binaryzation is obtained；
S6：Suspected target region is screened by Feature Selection device, positioning traffic sign region；Wherein, the spy
Levying screening washer is set up according to the shape facility of traffic sign.
According to one embodiment of present invention, the step S2 includes：
S21：Monochrome information is counted, dividing threshold value according to brightness divides luminance area, determines high-frequency region, intermediate frequency
Region and the pixel number of low frequency region, respectively with brightness statistics variable num_high, num_middle and num_low table
Show；
S22：The pixel number for calculating high-frequency region, mid-frequency region and low frequency region respectively is shared in all pixels point
Proportion P_high, P_middle and P_low,
Wherein,
M*N is all pixels point number；
S23：Image is divided into by dark scene, bright scene, backlight scene and normal scene according to scene classification threshold value, its
In,
Light scene is P_high ＞ 0.53＆P_low ＜ 0.35；
Dark scene is P_low ＞ 0.51；
Backlight scene is P_high+P_low ＞ 0.8＆P_low ＞ P_middle＆P_high ＞ P_middle；
Normal scene is other situations.
According to one embodiment of present invention, monochrome information is counted and is normalized, drawn according to brightness
Point threshold value divides luminance area：Low frequency region is [0,0.4]；Mid-frequency region is [0.4,0.7]；High-frequency region is [0.7,1].
According to one embodiment of present invention, in the step S3, algorithm is strengthened to classification chart picture using adaptive Gamma
Carry out image enhancement processing formula be：
F (I)=Iγ, wherein
Wherein, I is the gray value of each pixel of image of before processing, and f (I) is each pixel of image after processing
Gray value, γ is gamma parameters.
According to one embodiment of present invention, in the step S4, under RGB color, using improved three-component
Chromatism method, which carries out color segmentation, to be included：
The feature operator for extracting red area is λ R-G-B, and the R, G, B component of image are calculated according to formula (1) according to feature
Son carries out processing and extracts red area, obtains R component figure,
The feature operator for extracting blue region is into B-R-G, to the R, G, B component of image according to formula (2) according to feature
Operator carries out processing and extracts blue region, obtains B component figure,
Wherein, CA1 represents the grey scale pixel value of the red area extracted, and CA2 represents the pixel ash of the blue region extracted
Angle value, λ is adaptive weighted factor；
In the step S4, under RGB color, carrying out color segmentation using normalization RGB partitioning algorithms includes：
Calculate r=R/ (R+G+B), g=G/ (R+G+B), b=B/ (R+G+B)；
Pixel is red if r ＞=0.4＆g ＜ 0.3, and pixel is blueness if b ＞ 0.4, if r+g ＞ 0.85
Pixel is yellow.
According to one embodiment of present invention, the adaptive weighted factor is obtained as follows：
A1, the initial value that setting adaptive weighted factor enters；
A2, calculates the pixel number that pixel value on R component figure and B component figure is 0 according to formula (1) and (2) respectively, point
Do not represented with num_red0 and num_blue0；
A3, calculates target proportion k0=(num_red0+num_blue0)/2*M*N, M*N are all pixels point number；
A4, if judging k0＞ 0.98＆ λ≤2, then λ=λ+0.2, and return to step A2 are calculated；Otherwise, it determines now
Enter for final adaptive weighted factor.
According to one embodiment of present invention, the step S5 comprises the following steps：
S51：Image is divided into L gray level；
S52：For each gray level, it is that w0, average gray are uO that calculating prospect points, which account for image scaled, and background points are accounted for
Image scaled is that w1, average gray are u1；The overall average gray scale for calculating image is u=w0*u0+w1*u1, calculates foreground and background
The variance of image is g=w0* (u0-u) * (u0-u)+w1* (u1-u) * (u1-u)=w0*w1* (u0-u1) * (u0-u1)；
S53：Gray scale when variance g is maximum, is optimal segmenting threshold；
S54：The segmentation of prospect and background is carried out with the optimal segmenting threshold.
According to one embodiment of present invention, the step S5 includes：Carried out by the gray level image after color segmentation
After binary conversion treatment, using morphological erosion, expanding method processing is crude and plugs the gap, and obtains suspected target region.
According to one embodiment of present invention, in the step S6, using the shape facility of sign board, including length-width ratio,
Girth area ratio, region accounting invariant features similar with shape set up Feature Selection device, by Feature Selection device to suspected target
Region is screened, wherein：
Length-width ratio is 0.45 ＜ ratio1 ＜ 1.2；
Girth area ratio is 0.045 ＜ ratio2 ＜ 0.07 | ratio2 ＜ 0.038；
Region accounting is 0.58 ＜ ratio3 ＜ 0.63；And
Using the similar invariant features and space distribution information of shape contour, different criterions are set to different shape,
Profile coordinate is switched into polar coordinates (θ-ρ), triangle and rectangle circular according to the feature differentiation of ρ values.
According to one embodiment of present invention, in addition to step S7：Obtained traffic sign region and original image are carried out
Mask process：
S71：Using the bianry image in the traffic sign region of positioning as mask template, it is multiplied, is then handed over original image
Image value keeps constant in logical mark region, and the image value outside region is all 0；
S72：The image obtained to mask, statistical picture horizontal direction and the sum of vertical direction grey scale pixel value, are obtained respectively
To traffic sign region, then cut and extract.
After adopting the above technical scheme, the present invention has the advantages that compared with prior art：
Traffic mark board can be quickly detected for condition complicated under natural scene, and very high inspection can be reached
Survey rate, while meeting Mark Detection to robustness and requirement of real-time, proposes the scene classification method based on brightness case, solves
Influence of the complex illumination to RGB color component, overcomes and splits the not enough shortcoming of robustness based on rgb space hypograph；Knot
The screening split based on RGB color and based on shape facility has been closed, has realized by the thick Mark Detection to essence, inspection is greatly improved
Test the speed rate, it is ensured that Mark Detection requirement of real-time.
Brief description of the drawings
Fig. 1 for one embodiment of the invention natural scene under method for traffic sign detection schematic flow sheet；
Fig. 2 a are the image schematic diagram of the former detection image of one embodiment of the invention；
Fig. 2 b are the segmentation of a color image of one embodiment of the invention and obtain the image in the suspected target region after binaryzation
Schematic diagram；
Fig. 2 c are the screening of one embodiment of the invention and position the image schematic diagram behind traffic sign region；
Fig. 2 d are the image schematic diagram after the mask process of one embodiment of the invention；
The image schematic diagram in the traffic sign region that Fig. 2 e and 2f extract for the cutting of one embodiment of the invention.
Embodiment
In order to facilitate the understanding of the purposes, features and advantages of the present invention, below in conjunction with the accompanying drawings to the present invention
Embodiment be described in detail.
Many details are elaborated in the following description to fully understand the present invention.But the present invention can be with
Much it is different from other manner described here to implement, those skilled in the art can be in the situation without prejudice to intension of the present invention
Under do similar popularization, therefore the present invention is not limited to the specific embodiments disclosed below.
Referring to Fig. 1, in one embodiment, the method for traffic sign detection under natural scene comprises the following steps：
S1：Obtain the detection image shot under natural scene；
S2：Monochrome information to the detection image is counted, and different brightness regions are marked off according to grade brightness threshold value
Domain, calculates the pixel ratio of different luminance areas respectively, and image is divided into the moon according to each pixel ratio and scene classification threshold value
Dark scene, bright scene, backlight scene and normal scene；
S3：Gamma parameter values are chosen according to scene classification result, algorithm is strengthened to classification chart using adaptive Gamma
As carrying out image enhancement processing；
S4：Under RGB color, choose partitioning algorithm for different scenes and carry out segmentation of a color image, obtain doubtful
Target area；Wherein, for bright scene, color segmentation is carried out using normalization RGB partitioning algorithms；For dark scene, inverse
Light field scape and normal scene, carry out color segmentation, the improved three-component chromatism method using improved three-component chromatism method
By setting adaptive weighted factor come the adaptive color region for adjusting extraction；
S5：Gray level image after color segmentation is subjected to binary conversion treatment, the suspected target region after binaryzation is obtained；
S6：Suspected target region is screened by Feature Selection device, positioning traffic sign region；Wherein, the spy
Levying screening washer is set up according to the shape facility of traffic sign.
The method for traffic sign detection under the natural scene of the embodiment of the present invention is specifically retouched below in conjunction with the accompanying drawings
State, but should be as limit.
In step sl, can be by the captured in real-time during stroke, to obtain the detection figure shot under natural scene
Picture, certainly, the detection image not shot every time occurs can be with the presence or absence of traffic sign in traffic sign, detection image
Detected by follow-up step, the image can be neglected if without traffic sign.In the process step of the present embodiment,
There is traffic sign in detection image.Certainly, due to subsequently needing to carry out image procossing, thus image in rgb space
Shoot to be coloured image, Fig. 2 a eliminate its color for the ease of description.
Then step S2 is performed, the monochrome information of the detection image to being obtained in step S1 is counted, it is bright according to grade
Degree threshold value marks off different luminance areas, the pixel ratio of different luminance areas is calculated respectively, according to each pixel ratio and scene
Classification thresholds and image is divided into dark scene, bright scene, backlight scene and normal scene.
Can be by brightness histogram, the notable difference scope of brightness section, reduces classification under relatively more different light conditions
Interval, then for interval debugging so as to adjust scene classification threshold value, to obtain optimal scene classification result.
Further, step S2 can include：
S21：Monochrome information is counted, dividing threshold value according to brightness divides luminance area, determines high-frequency region, intermediate frequency
Region and the pixel number of low frequency region, respectively with brightness statistics variable num_high, num_middle and num_low table
Show；
S22：The pixel number for calculating high-frequency region, mid-frequency region and low frequency region respectively is shared in all pixels point
Proportion P_high, P_middle and P_low,
Wherein,M*
N is all pixels point number；
S23：Image is divided into by dark scene, bright scene, backlight scene and normal scene according to scene classification threshold value, its
In,
Light scene is P_high ＞ 0.53＆P_low ＜ 0.35；
Dark scene is P_low ＞ 0.51；
Backlight scene is P_high+P_low ＞ 0.8＆P_low ＞ P_middle＆P_high ＞ P_middle；
Normal scene is other situations.
Optionally, monochrome information is counted and is normalized, dividing threshold value according to brightness divides brightness region
Domain is：Low frequency region is [0,0.4]；Mid-frequency region is [0.4,0.7]；High-frequency region is [0.7,1].
Then step S3 is performed, gamma parameter values are chosen according to scene classification result, is strengthened using adaptive Gamma
Algorithm carries out image enhancement processing to classification chart picture.
Specifically, in step S3, the public affairs that algorithm carries out image enhancement processing to classification chart picture are strengthened using adaptive Gamma
Formula is：
F (I)=Iγ, wherein
Wherein, I is the gray value of each pixel of image of before processing, and f (I) is each pixel of image after processing
Gray value, γ is gamma parameters.
According to scene classification result, classification chart picture is carried out at respective image enhancing using adaptive Gamma Enhancement Methods
Reason, is converted different from general Gamma, and the value of gamma parameters is adaptively chosen according to classification results in the present embodiment, i.e., bright
Light field scape is 0.5；Dark scene is 2.2；Backlight scene is 1.2；Normal scene is 1.
RGB color is very sensitive to illumination variation, easily causes missing inspection or false retrieval situation, by brightness scene classification simultaneously
The enhancing processing of adaptability is carried out, influence of the illumination to RGB color is eliminated, it is to avoid the problem of missing inspection or false retrieval, gram
Take and the not enough shortcoming of robustness is split based on rgb space hypograph.
Then step S4 is performed, is carried out based on the mark coarse positioning under rgb space.Under RGB color, for difference
Scene chooses partitioning algorithm and carries out segmentation of a color image, obtains suspected target region；Wherein, for bright scene, using normalizing
Change RGB partitioning algorithms and carry out color segmentation；For dark scene, backlight scene and normal scene, using improved three-component
Chromatism method carries out color segmentation, and the improved three-component chromatism method is by setting adaptive weighted factor to be carried come adaptive adjust
The color region taken.
Because brightness scene classification eliminates influence of the illumination to RGB color, therefore can be directly under rgb space
Color segmentation is carried out to enhanced image.According to brightness scene classification results, adaptively selected improved three-component chromatism method
With normalization RGB split plot designs.
Specifically, in step s 4, under RGB color, color segmentation is carried out using improved three-component chromatism method
Including：
The feature operator for extracting red area is into R-G-B, to the R, G, B component of image according to formula (1) according to feature
Operator carries out processing and extracts red area, obtains R component figure,
The feature operator for extracting blue region is λ B-R-G, and the R, G, B component of image are calculated according to formula (2) according to feature
Son carries out processing and extracts blue region, obtains B component figure,
Wherein, CA1 represents the grey scale pixel value of the red area extracted, and CA2 represents the pixel ash of the blue region extracted
Angle value, λ is adaptive weighted factor.
Because red threshold interval and yellow is close, while extracting red, yellow can also be obtained, it is only necessary to
RGB component can be extracted by extracting red area and extracting blue region.
Weighted factor in the past is definite value, without robustness；And weighted factor is improved, can be according to difference
Image carries out automatic adjusument.If weighted factor is definite value 1.6, dark image is likely to occur missing inspection, and backlight image occurs
Cross inspection；It can improve this phenomenon after adaptive adjustment.
In step s 4, under RGB color, carrying out color segmentation using normalization RGB partitioning algorithms includes：
Calculate r=R/ (R+G+B), g=G/ (R+G+B), b=B/ (R+G+B)；
Pixel is red if r ＞=0.4＆g ＜ 0.3, and pixel is blueness if b ＞ 0.4, if r+g ＞ 0.85
Pixel is yellow.R, g, b are interim scale parameter.
Although brightness scene classification reduces influence of the illumination to rgb space, change for direct use of bright scene
During the three-component chromatism method entered, imperfect situation occurs in the image processing edge extracted, and uses normalization RGB split plot designs
When avoid such case, thus bright scene is used into the color segmentation method different from other scenes.
It is preferred that, adaptive weighted factor can be obtained as follows：
A1, the initial value that setting adaptive weighted factor enters；Initial value could be arranged to 1.2；
A2, calculates the pixel number that pixel value on R component figure and B component figure is 0 according to formula (1) and (2) respectively, point
Do not represented with num_red0 and num_blue0；The pixel number that component value needed for image is 0 is namely calculated respectively；
A3, calculates target proportion k0=(num_red0+num_blue0)/2*M*N, M*N are all pixels point number；
A4, if judging k0＞ 0.98＆ λ≤2, then λ=λ+0.2, and return to step A2 are calculated；Otherwise, it determines now
λ be final adaptive weighted factor.
The embodiment of the present invention realizes brightness scene classification according to brightness value, is divided into picture based on classificating thought different bright
Scene is spent, the extraction algorithm then matched for Sexual behavior mode improves robustness；If without scene classification, and with same
Method, then without robustness, i.e., extraction algorithm may be only suitable for normal illumination situation, at the weak or strong image of illumination
Manage effect bad.
Then step S5 is performed, the gray level image after color segmentation is subjected to binary conversion treatment, obtained after binaryzation
Suspected target region.Occur that the obtained image behind suspicious region, but segmentation is gray level image after segmentation, it is therefore desirable to carry out
Binary conversion treatment obtains bianry image.
Specifically, step S5 comprises the following steps：
S51：Image is divided into L gray level；
S52：For each gray level, it is that w0, average gray are u0 that calculating prospect points, which account for image scaled, and background points are accounted for
Image scaled is that w1, average gray are u1；The overall average gray scale for calculating image is u=w0*u0+w1*u1, calculates foreground and background
The variance of image is g=w0* (u0-u) * (u0-u)+w1* (u1-u) * (u1-u)=w0*w1* (u0-u1) * (u0-u1)；
S53：Gray scale when variance g is maximum, is optimal segmenting threshold；
S54：The segmentation of prospect and background is carried out with the optimal segmenting threshold.
When variance g is maximum, it is believed that now foreground and background difference is maximum, and gray scale t now is optimal threshold, figure
As upper gray value less than t is all changed into 0, all changes 1 higher than t.
It is preferred that, step S5 includes：After the gray level image after color segmentation to be carried out to binary conversion treatment, shape is used
The processing of state burn into expanding method is crude and plugs the gap, and obtains suspected target region.Suspected target region obtain after figure
As shown in Figure 2 b.
Then step S6 is performed, suspected target region is screened by Feature Selection device, positioning traffic sign region；
Wherein, the Feature Selection device is set up according to the shape facility of traffic sign.
For the suspected target region extracted in step S5, non-mark region is still suffered from, it is therefore desirable to remove the part
Suspicious region.Using Feature Selection device, extraneous areas is removed, being accurately positioned for traffic sign is realized, obtains traffic sign region.
Specifically, in step S6, using the shape facility of sign board, including length-width ratio, girth area ratio, region accounting and
The similar invariant features of shape, set up Feature Selection device, and suspected target region is screened by Feature Selection device, wherein：
Length-width ratio is 0.45 ＜ ratio1 ＜ 1.2；
Girth area ratio is 0.045 ＜ ratio2 ＜ 0.07 | ratio2 ＜ 0.038；
Region accounting is 0.58 ＜ ratio3 ＜ 0.63；And
Using the similar invariant features and space distribution information of shape contour, different criterions are set to different shape,
Profile coordinate is switched into polar coordinates (θ-ρ), triangle and rectangle circular according to the feature differentiation of ρ values.
Using the similar invariant features of profile, the problem of can solving to noise-sensitives such as image translation, rotation, contractions.Essence
It is determined that behind position, obtaining traffic sign region as shown in Figure 2 c.
In one embodiment, the method for traffic sign detection under natural scene can also include step S7：By what is obtained
Traffic sign region carries out mask process with original image：
S71：Using the bianry image in the traffic sign region of positioning as mask template, it is multiplied, is then handed over original image
Image value keeps constant in logical mark region, and the image value outside region is all 0；Image after mask process is as shown in Figure 2 d；
S72：The image obtained to mask, statistical picture horizontal direction and the sum of vertical direction grey scale pixel value, are obtained respectively
To traffic sign region, then cut and extract.Cut the image extracted as shown in figure 2 e and 2f.
The present invention carries out checking test in German traffic sign database (GTSDB), test result indicates that this method is simultaneous
Turn round and look at robustness and requirement of real-time.Compared with current road traffic sign detection identification field related data, the present invention uses luminance field
Scape classification overcomes influence of the illumination to RGB component, and reach with the marker detection method being combined based on RGB and shape facility
To compared with high detection rates, showing preferable robustness and real-time.
Although the present invention is disclosed as above with preferred embodiment, it is not for limiting claim, any this area
Technical staff without departing from the spirit and scope of the present invention, can make possible variation and modification, therefore the present invention
The scope that protection domain should be defined by the claims in the present invention is defined.
Claims (10)
1. the method for traffic sign detection under a kind of natural scene, it is characterised in that comprise the following steps：
S1：Obtain the detection image shot under natural scene；
S2：Monochrome information to the detection image is counted, and different luminance areas are marked off according to grade brightness threshold value, point
The pixel ratio of different luminance areas is not calculated, and image is divided into by dark field according to each pixel ratio and scene classification threshold value
Scape, bright scene, backlight scene and normal scene；
S3：Gamma parameter values are chosen according to scene classification result, classification chart picture entered using adaptive Gamma enhancing algorithms
Row image enhancement processing；
S4：Under RGB color, choose partitioning algorithm for different scenes and carry out segmentation of a color image, obtain suspected target
Region；Wherein, for bright scene, color segmentation is carried out using normalization RGB partitioning algorithms；For dark scene, backlight
Scape and normal scene, carry out color segmentation using improved three-component chromatism method, and the improved three-component chromatism method passes through
The adaptive color region for adjusting extraction of adaptive weighted factor is set；
S5：Gray level image after color segmentation is subjected to binary conversion treatment, the suspected target region after binaryzation is obtained；
S6：Suspected target region is screened by Feature Selection device, positioning traffic sign region；Wherein, the feature sieve
Selecting device is set up according to the shape facility of traffic sign.
2. the method for traffic sign detection under natural scene as claimed in claim 1, it is characterised in that the step S2 bags
Include：
S21：Monochrome information is counted, dividing threshold value according to brightness divides luminance area, determines high-frequency region, mid-frequency region
With the pixel number of low frequency region, represented respectively with brightness statistics variable num_high, num_middle and num_low；
S22：The pixel number of high-frequency region, the mid-frequency region and low frequency region proportion in all pixels point is calculated respectively
P_high, P_middle and P_low,
Wherein,
M*N is all pixels point number；
S23：Image is divided into by dark scene, bright scene, backlight scene and normal scene according to scene classification threshold value, wherein,
Light scene is P_high>0.53&P_low<0.35；
Dark scene is P_low>0.51；
Backlight scene is P_high+P_low>0.8&P_low>P_middle&P_high>P_middle；
Normal scene is other situations.
3. the method for traffic sign detection under natural scene as claimed in claim 2, it is characterised in that carried out to monochrome information
Count and be normalized, dividing threshold value division luminance area according to brightness is：Low frequency region is [0,0.4]；Intermediate frequency zone
Domain is [0.4,0.7]；High-frequency region is [0.7,1].
4. the method for traffic sign detection under natural scene as claimed in claim 1, it is characterised in that in the step S3,
Use adaptive Gamma enhancing algorithm classification chart picture is carried out the formula of image enhancement processing for：
F (I)=Iγ, wherein
Wherein, I is the gray value of each pixel of image of before processing, and f (I) is the gray scale of each pixel of image after processing
Value, γ is gamma parameters.
5. the method for traffic sign detection under natural scene as claimed in claim 1, it is characterised in that in the step S4,
Under RGB color, carrying out color segmentation using improved three-component chromatism method includes：
The feature operator for extracting red area is λ R-G-B, and the R, G, B component of image are entered according to formula (1) according to feature operator
Red area is extracted in row processing, obtains R component figure,
The feature operator for extracting blue region is λ B-R-G, and the R, G, B component of image are entered according to formula (2) according to feature operator
Blue region is extracted in row processing, obtains B component figure,
Wherein, CA1 represents the grey scale pixel value of the red area extracted, and CA2 represents the grey scale pixel value of the blue region extracted,
λ is adaptive weighted factor；
In the step S4, under RGB color, carrying out color segmentation using normalization RGB partitioning algorithms includes：
Calculate r=R/ (R+G+B), g=G/ (R+G+B), b=B/ (R+G+B)；
If r>=0.4＆g<0.3 pixel is red, if b>0.4 pixel is blueness, if r+g>0.85 pixel is
Yellow.
6. the method for traffic sign detection under natural scene as claimed in claim 5, it is characterised in that described adaptive weighted
The factor is obtained as follows：
A1, setting adaptive weighted factor λ initial value；
A2, calculates the pixel number that pixel value on R component figure and B component figure is 0 according to formula (1) and (2) respectively, uses respectively
Num_red0 and num_blue0 are represented；
A3, calculates target proportion k0=(num_red0+num_blue0)/2*M*N, M*N are all pixels point number；
A4, if judging k0＞ 0.98＆ λ≤2, then λ=λ+0.2, and return to step A2 are calculated；Otherwise, it determines λ now is
Final adaptive weighted factor.
7. the method for traffic sign detection under natural scene as claimed in claim 1, it is characterised in that the step S5 includes
Following steps：
S51：Image is divided into L gray level；
S52：For each gray level, it is that w0, average gray are u0 that calculating prospect points, which account for image scaled, and background points account for image
Ratio is that w1, average gray are u1；The overall average gray scale for calculating image is u=w0*u0+w1*u1, calculates foreground and background image
Variance be g=w0* (u0-u) * (u0-u)+w1* (u1-u) * (u1-u)=w0*w1* (u0-u1) * (u0-u1)；
S53：Gray scale when variance g is maximum, is optimal segmenting threshold；
S54：The segmentation of prospect and background is carried out with the optimal segmenting threshold.
8. the method for traffic sign detection under natural scene as described in claim 1 or 7, it is characterised in that the step S5
Including：After the gray level image after color segmentation to be carried out to binary conversion treatment, morphological erosion, expanding method processing are used
It is crude and plug the gap, obtain suspected target region.
9. the method for traffic sign detection under natural scene as claimed in claim 1, it is characterised in that in the step S6,
Feature is set up using the shape facility of sign board, including length-width ratio, girth area ratio, region accounting invariant features similar with shape
Screening washer, is screened by Feature Selection device to suspected target region, wherein：
Length-width ratio is 0.45<ratio1<1.2；
Girth area ratio is 0.045<ratio2<0.07|ratio2<0.038；
Region accounting is 0.58<ratio3<0.63；And
Using the similar invariant features and space distribution information of shape contour, different criterions are set to different shape, will be taken turns
Wide coordinate switchs to polar coordinates (θ-ρ), circular according to the feature differentiation of ρ values, triangle and rectangle.
10. the method for traffic sign detection under natural scene as claimed in claim 1, it is characterised in that also including step S7：
Obtained traffic sign region and original image are subjected to mask process：
S71：Using the bianry image in the traffic sign region of positioning as mask template, it is multiplied with original image, then traffic mark
Image value keeps constant in will region, and the image value outside region is all 0；
S72：The image obtained to mask, statistical picture horizontal direction and the sum of vertical direction grey scale pixel value, are handed over respectively
Logical mark region, then cuts and extracts.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710540228.3A CN107301405A (en) | 2017-07-04 | 2017-07-04 | Method for traffic sign detection under natural scene |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710540228.3A CN107301405A (en) | 2017-07-04 | 2017-07-04 | Method for traffic sign detection under natural scene |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107301405A true CN107301405A (en) | 2017-10-27 |
Family
ID=60136146
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710540228.3A Pending CN107301405A (en) | 2017-07-04 | 2017-07-04 | Method for traffic sign detection under natural scene |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107301405A (en) |
Cited By (28)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108024105A (en) * | 2017-12-14 | 2018-05-11 | 珠海市君天电子科技有限公司 | Image color adjusting method, device, electronic equipment and storage medium |
| CN108399610A (en) * | 2018-03-20 | 2018-08-14 | 上海应用技术大学 | A kind of depth image enhancement method of fusion RGB image information |
| CN108711160A (en) * | 2018-05-18 | 2018-10-26 | 西南石油大学 | A kind of Target Segmentation method based on HSI enhancement models |
| CN108734131A (en) * | 2018-05-22 | 2018-11-02 | 杭州电子科技大学 | A kind of traffic sign symmetry detection methods in image |
| CN108765443A (en) * | 2018-05-22 | 2018-11-06 | 杭州电子科技大学 | A kind of mark enhancing processing method of adaptive color Threshold segmentation |
| CN109214434A (en) * | 2018-08-20 | 2019-01-15 | 上海萃舟智能科技有限公司 | A kind of method for traffic sign detection and device |
| CN109815906A (en) * | 2019-01-25 | 2019-05-28 | 华中科技大学 | Method for traffic sign detection and system based on substep deep learning |
| CN109916415A (en) * | 2019-04-12 | 2019-06-21 | 北京百度网讯科技有限公司 | Road type determines method, apparatus, equipment and storage medium |
| CN110312106A (en) * | 2019-07-30 | 2019-10-08 | 广汽蔚来新能源汽车科技有限公司 | Display methods, device, computer equipment and the storage medium of image |
| CN110598705A (en) * | 2019-09-27 | 2019-12-20 | 腾讯科技（深圳）有限公司 | Semantic annotation method and device for image |
| CN110619648A (en) * | 2019-09-19 | 2019-12-27 | 四川长虹电器股份有限公司 | Method for dividing image area based on RGB change trend |
| CN110838131A (en) * | 2019-11-04 | 2020-02-25 | 网易（杭州）网络有限公司 | Method and device for realizing automatic cutout, electronic equipment and medium |
| CN110930358A (en) * | 2019-10-17 | 2020-03-27 | 广州丰石科技有限公司 | Solar panel image processing method based on self-adaptive algorithm |
| CN111062309A (en) * | 2019-12-13 | 2020-04-24 | 吉林大学 | Method, storage medium and system for detecting traffic signs in rainy days |
| CN111275648A (en) * | 2020-01-21 | 2020-06-12 | 腾讯科技（深圳）有限公司 | Face image processing method, device and equipment and computer readable storage medium |
| CN111369634A (en) * | 2020-03-26 | 2020-07-03 | 苏州瑞立思科技有限公司 | Image compression method and device based on weather conditions |
| 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 |
| CN111723805A (en) * | 2019-03-18 | 2020-09-29 | 浙江宇视科技有限公司 | Signal lamp foreground area identification method and related device |
| CN111860533A (en) * | 2019-04-30 | 2020-10-30 | 深圳数字生命研究院 | Image recognition method and device, storage medium and electronic device |
| CN112052700A (en) * | 2019-06-06 | 2020-12-08 | 北京京东尚科信息技术有限公司 | Image binarization threshold matrix determination and graphic code information identification method and device |
| CN112226812A (en) * | 2020-10-20 | 2021-01-15 | 北京图知天下科技有限责任公司 | Czochralski monocrystalline silicon production method, device and system |
| CN112507911A (en) * | 2020-12-15 | 2021-03-16 | 浙江科技学院 | Real-time recognition method of pecan fruits in image based on machine vision |
| CN112699841A (en) * | 2021-01-13 | 2021-04-23 | 华南理工大学 | Traffic sign detection and identification method based on driving video |
| CN112906712A (en) * | 2021-03-02 | 2021-06-04 | 湖南金烽信息科技有限公司 | Neural network image preprocessing method based on light intensity analysis |
| US11044410B2 (en) | 2018-08-13 | 2021-06-22 | Guangdong Oppo Mobile Telecommunications Corp., Ltd. | Imaging control method and apparatus, electronic device, and computer readable storage medium |
| CN113610185A (en) * | 2021-08-19 | 2021-11-05 | 江西应用技术职业学院 | Wood color sorting method based on dominant hue identification |
| CN114511770A (en) * | 2021-12-21 | 2022-05-17 | 武汉光谷卓越科技股份有限公司 | Road sign plate identification method |
| CN114863109A (en) * | 2022-05-25 | 2022-08-05 | 广东飞达交通工程有限公司 | Segmentation technology-based fine recognition method for various targets and elements of traffic scene |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105787475A (en) * | 2016-03-29 | 2016-07-20 | 西南交通大学 | Traffic sign detection and identification method under complex environment |
| CN106503704A (en) * | 2016-10-21 | 2017-03-15 | 河南大学 | Circular traffic sign localization method in a kind of natural scene |
-
2017
- 2017-07-04 CN CN201710540228.3A patent/CN107301405A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105787475A (en) * | 2016-03-29 | 2016-07-20 | 西南交通大学 | Traffic sign detection and identification method under complex environment |
| CN106503704A (en) * | 2016-10-21 | 2017-03-15 | 河南大学 | Circular traffic sign localization method in a kind of natural scene |
Non-Patent Citations (1)
| Title |
|---|
| 任敬义: "自然场景中交通标志的检测与识别", 《中国优秀硕士学位论文全文数据库 工程科技II辑》 * |
Cited By (44)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108024105A (en) * | 2017-12-14 | 2018-05-11 | 珠海市君天电子科技有限公司 | Image color adjusting method, device, electronic equipment and storage medium |
| CN108399610A (en) * | 2018-03-20 | 2018-08-14 | 上海应用技术大学 | A kind of depth image enhancement method of fusion RGB image information |
| CN108711160A (en) * | 2018-05-18 | 2018-10-26 | 西南石油大学 | A kind of Target Segmentation method based on HSI enhancement models |
| CN108711160B (en) * | 2018-05-18 | 2022-06-14 | 西南石油大学 | Target segmentation method based on HSI (high speed input/output) enhanced model |
| CN108765443A (en) * | 2018-05-22 | 2018-11-06 | 杭州电子科技大学 | A kind of mark enhancing processing method of adaptive color Threshold segmentation |
| CN108765443B (en) * | 2018-05-22 | 2021-08-24 | 杭州电子科技大学 | Sign enhancement processing method for self-adaptive color threshold segmentation |
| CN108734131A (en) * | 2018-05-22 | 2018-11-02 | 杭州电子科技大学 | A kind of traffic sign symmetry detection methods in image |
| CN108734131B (en) * | 2018-05-22 | 2021-08-17 | 杭州电子科技大学 | Method for detecting symmetry of traffic sign in image |
| US11044410B2 (en) | 2018-08-13 | 2021-06-22 | Guangdong Oppo Mobile Telecommunications Corp., Ltd. | Imaging control method and apparatus, electronic device, and computer readable storage medium |
| US11765466B2 (en) | 2018-08-13 | 2023-09-19 | Guangdong Oppo Mobile Telecommunications Corp., Ltd. | Imaging control method and apparatus, electronic device, and computer readable storage medium |
| CN109214434A (en) * | 2018-08-20 | 2019-01-15 | 上海萃舟智能科技有限公司 | A kind of method for traffic sign detection and device |
| CN109815906A (en) * | 2019-01-25 | 2019-05-28 | 华中科技大学 | Method for traffic sign detection and system based on substep deep learning |
| CN109815906B (en) * | 2019-01-25 | 2021-04-06 | 华中科技大学 | Traffic sign detection method and system based on step-by-step deep learning |
| CN111723805B (en) * | 2019-03-18 | 2023-06-20 | 浙江宇视科技有限公司 | Method and related device for identifying foreground region of signal lamp |
| CN111723805A (en) * | 2019-03-18 | 2020-09-29 | 浙江宇视科技有限公司 | Signal lamp foreground area identification method and related device |
| CN109916415A (en) * | 2019-04-12 | 2019-06-21 | 北京百度网讯科技有限公司 | Road type determines method, apparatus, equipment and storage medium |
| CN111860533A (en) * | 2019-04-30 | 2020-10-30 | 深圳数字生命研究院 | Image recognition method and device, storage medium and electronic device |
| CN111860533B (en) * | 2019-04-30 | 2023-12-12 | 深圳数字生命研究院 | Image recognition method and device, storage medium and electronic device |
| CN112052700B (en) * | 2019-06-06 | 2024-04-05 | 北京京东乾石科技有限公司 | Image binarization threshold matrix determination and graphic code information identification method and device |
| CN112052700A (en) * | 2019-06-06 | 2020-12-08 | 北京京东尚科信息技术有限公司 | Image binarization threshold matrix determination and graphic code information identification method and device |
| CN110312106B (en) * | 2019-07-30 | 2021-05-18 | 广汽蔚来新能源汽车科技有限公司 | Image display method and device, computer equipment and storage medium |
| CN110312106A (en) * | 2019-07-30 | 2019-10-08 | 广汽蔚来新能源汽车科技有限公司 | Display methods, device, computer equipment and the storage medium of image |
| CN110619648A (en) * | 2019-09-19 | 2019-12-27 | 四川长虹电器股份有限公司 | Method for dividing image area based on RGB change trend |
| CN110619648B (en) * | 2019-09-19 | 2022-03-15 | 四川长虹电器股份有限公司 | Method for dividing image area based on RGB change trend |
| CN110598705A (en) * | 2019-09-27 | 2019-12-20 | 腾讯科技（深圳）有限公司 | Semantic annotation method and device for image |
| CN110930358B (en) * | 2019-10-17 | 2023-04-21 | 广州丰石科技有限公司 | Solar panel image processing method based on self-adaptive algorithm |
| CN110930358A (en) * | 2019-10-17 | 2020-03-27 | 广州丰石科技有限公司 | Solar panel image processing method based on self-adaptive algorithm |
| CN110838131A (en) * | 2019-11-04 | 2020-02-25 | 网易（杭州）网络有限公司 | Method and device for realizing automatic cutout, electronic equipment and medium |
| CN110838131B (en) * | 2019-11-04 | 2022-05-17 | 网易（杭州）网络有限公司 | Method and device for realizing automatic cutout, electronic equipment and medium |
| CN111062309A (en) * | 2019-12-13 | 2020-04-24 | 吉林大学 | Method, storage medium and system for detecting traffic signs in rainy days |
| CN111062309B (en) * | 2019-12-13 | 2022-12-30 | 吉林大学 | Method, storage medium and system for detecting traffic signs in rainy days |
| CN111275648B (en) * | 2020-01-21 | 2024-02-09 | 腾讯科技（深圳）有限公司 | Face image processing method, device, equipment and computer readable storage medium |
| CN111275648A (en) * | 2020-01-21 | 2020-06-12 | 腾讯科技（深圳）有限公司 | Face image processing method, device and equipment and computer readable storage medium |
| CN111369634A (en) * | 2020-03-26 | 2020-07-03 | 苏州瑞立思科技有限公司 | Image compression method and device based on weather conditions |
| 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 |
| CN112226812A (en) * | 2020-10-20 | 2021-01-15 | 北京图知天下科技有限责任公司 | Czochralski monocrystalline silicon production method, device and system |
| CN112507911A (en) * | 2020-12-15 | 2021-03-16 | 浙江科技学院 | Real-time recognition method of pecan fruits in image based on machine vision |
| CN112699841A (en) * | 2021-01-13 | 2021-04-23 | 华南理工大学 | Traffic sign detection and identification method based on driving video |
| CN112906712A (en) * | 2021-03-02 | 2021-06-04 | 湖南金烽信息科技有限公司 | Neural network image preprocessing method based on light intensity analysis |
| CN113610185B (en) * | 2021-08-19 | 2022-03-22 | 江西应用技术职业学院 | Wood color sorting method based on dominant hue identification |
| CN113610185A (en) * | 2021-08-19 | 2021-11-05 | 江西应用技术职业学院 | Wood color sorting method based on dominant hue identification |
| CN114511770A (en) * | 2021-12-21 | 2022-05-17 | 武汉光谷卓越科技股份有限公司 | Road sign plate identification method |
| CN114863109A (en) * | 2022-05-25 | 2022-08-05 | 广东飞达交通工程有限公司 | Segmentation technology-based fine recognition method for various targets and elements of traffic scene |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107301405A (en) | Method for traffic sign detection under natural scene | |
| CN103761529B (en) | A kind of naked light detection method and system based on multicolour model and rectangular characteristic | |
| CN105205489B (en) | Detection method of license plate based on color and vein analyzer and machine learning | |
| CN105894701B (en) | The identification alarm method of transmission line of electricity external force damage prevention Large Construction vehicle | |
| CN105160297B (en) | Masked man's event automatic detection method based on features of skin colors | |
| Singh et al. | Local contrast and mean based thresholding technique in image binarization | |
| Yu et al. | A classification algorithm to distinguish image as haze or non-haze | |
| CN109086687A (en) | The traffic sign recognition method of HOG-MBLBP fusion feature based on PCA dimensionality reduction | |
| CN107705254A (en) | A kind of urban environment appraisal procedure based on streetscape figure | |
| CN112818853B (en) | Traffic element identification method, device, equipment and storage medium | |
| CN102855627B (en) | City remote sensing image shadow detection method based on spectral characteristic and topological relation | |
| CN111539980B (en) | Multi-target tracking method based on visible light | |
| Chen et al. | Robust license plate detection in nighttime scenes using multiple intensity IR-illuminator | |
| CN112017445A (en) | Pedestrian violation prediction and motion trail tracking system and method | |
| Avery et al. | Investigation into shadow removal from traffic images | |
| Wang et al. | Adaboost-based crack detection method for pavement | |
| CN107122732A (en) | The quick license plate locating method of high robust under a kind of monitoring scene | |
| CN110097524A (en) | SAR image object detection method based on fusion convolutional neural networks | |
| Jia et al. | Design of Traffic Sign Detection and Recognition Algorithm Based on Template Matching | |
| CN106650824A (en) | Moving object classification method based on support vector machine | |
| CN113221603A (en) | Method and device for detecting shielding of monitoring equipment by foreign matters | |
| CN111666811A (en) | Method and system for extracting traffic sign area in traffic scene image | |
| Lafuente-Arroyo et al. | Traffic sign classification invariant to rotations using support vector machines | |
| CN110633705A (en) | Low-illumination imaging license plate recognition method and device | |
| Patel et al. | A novel approach for detecting number plate based on overlapping window and region clustering for Indian conditions |
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: 20171027 |