CN104537350A - License plate positioning method based on multi-dimensional edge feature - Google Patents
License plate positioning method based on multi-dimensional edge feature Download PDFInfo
- Publication number
- CN104537350A CN104537350A CN201410838603.9A CN201410838603A CN104537350A CN 104537350 A CN104537350 A CN 104537350A CN 201410838603 A CN201410838603 A CN 201410838603A CN 104537350 A CN104537350 A CN 104537350A
- Authority
- CN
- China
- Prior art keywords
- image
- license plate
- gray
- pixel
- edge
- 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
- G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a license plate positioning method based on a multi-dimensional edge feature. The method comprises the steps that firstly, an original image is processed through a related image pre-processing method; secondly, edge points not conforming to conditions are removed through image edge length and edge density scanning; thirdly, the original image is processed through a color descending dimension method, and color feature information corresponding to edge feature points left in the second step is obtained; fourthly, a license plate area is accurately positioned according to the color information of the image edge feature points. The license plate positioning method is used for acquiring segmenting information of a reference image and is applied to a license plate positioning system, the calculation speed is higher, the real-time performance is good, the positioning information is accurate, and the capacity for resisting interference of the background of the image is high.
Description
Technical field
The present invention is mainly concerned with the disposal route of image, particularly relates to the license plate locating method based on multiple edge features in a kind of Position System of automobile license plate location.
Background technology
License plate recognition technology all has a wide range of applications in the intelligent traffic administration systems such as wagon flow monitoring, electronic charging, and License Plate is a gordian technique of Vehicle License Plate Recognition System, whether the accurate positioning of license plate area, to directly have influence on the Recognition of License Plate Characters efficiency in later stage, thus have influence on the efficiency of whole system.According to the soft and hardware condition of different study period, license plate locating method is broadly divided into the license plate locating method based on coloured image and the license plate locating method based on gray level image.A kind of front method remains all information of license plate image, adds the complexity of algorithm, although the degree of accuracy of location is improved, but real-time is not strong.Then a kind of method processes based on gray level image, combines, achieve certain achievement by traditional image processing method and vehicle license plate characteristic.The localization method reaction time wherein based on rim detection is fast, and locating accuracy is high, effectively can remove noise, and it is also use one of method the most widely in vehicle license location technique field.
In the rim detection vehicle license location technique of existing relative maturity, as the algorithm of locating license plate of vehicle (License plate location based on multiple edge features) based on multiple edge feature, its main thought is the colouring information feature of marginal information and the license plate area enriched in conjunction with license plate area, accurately locates from candidate license plate region.Experimental result shows, the method accuracy rate is high, strong robustness.But this algorithm is less relative to background to license plate image, car plate knockdown when, locating effect is very micro-.Car plate colouring information in addition owing to applying in this algorithm does not do a lot of simplify processes, and therefore need amount of image information to be processed large, the complexity of algorithm is high, the overlong time of required process, and real-time is not strong.
" algorithm of locating license plate of vehicle based on gray variance and marginal density " (the Carlicense plate location algorithm based on intensity variance and edgedensity) of Zhang Haopeng and Wang Zongyi, according to license plate area, there is gray variance approximately equal and the approximately equalised character of marginal density within the specific limits, propose the matched filter based on license plate area marginal density character, this wave filter can extract all candidate targets effectively.The advantage of this algorithm is that it effectively can improve the picture quality of license plate area and the accuracy rate of car plate target localization, has certain practical value.But the method to environment as the change of the conditions such as illumination, visual angle, Distance geometry background the unfavorable factor that causes there is no good inhibiting effect, thus also reduce its location efficiency.
For solving the problems referred to above existed in License Plate, the invention provides a kind of newly, effective solution.
Summary of the invention
For making up the deficiencies in the prior art, the present invention seeks to be to propose a kind of speed faster, in the higher Position System of automobile license plate location of locating accuracy based on the license plate locating method of multiple edge features, avoid the uncertainty of binary-state threshold value, decrease the calculated amount of algorithm of locating license plate of vehicle, have certain anti-interference to the unfavorable factor that the change of the conditions such as environment such as illumination, visual angle, Distance geometry background causes.
For achieving the above object, the present invention adopts following technical scheme: a kind of license plate locating method based on multi-dimensional edge feature, and concrete steps are as follows:
The first step, with the relevant former license plate image of image pre-processing method process, due to the vertical edge abundant information of license plate area, therefore used image processing method makes the vertical edge information displaying of image out as far as possible;
Second step, utilizes the car plate edge feature of regulation, scans pretreated image, reject and be not inconsistent the image border point of standard, thus the partial noise in filtering image.In the image obtained, similar license plate area is as the candidate license plate region of the 4th step;
3rd step, adopts color dimension reduction method to process initial pictures, draws in second step and remain the corresponding colouring information feature of Edge Feature Points;
4th step, the colouring information of combining image Edge Feature Points, finally accurately orients license plate area.
In the described first step, because the information that RGB color image comprises is too much, need first to carry out pre-service to image, obtain the image-region of marginal information compared with horn of plenty.Its concrete steps are: RGB image is converted to gray level image, carry out grey level enhancement to the gray level image obtained.
To all pixel p=(x in gray level image
p, y
p) in its neighborhood window W, ask gray variance respectively
average gray value
and average gray variance Best.By I
pwith I'
pbe denoted as the pixel value before and after pixel p grey level enhancement respectively.Its enhancing function
and
meaning be exactly gray variance near Best time amplify; Vertical edge operator is utilized to carry out rim detection to gray level image.Major part license plate area has common marginal information feature, i.e. vertical edge abundant information.Vertical edge detection is carried out to the image after grey level enhancement, f (x, y)=BYTE (abs (f (x, y-1)-f (x is used to all pixel p, y+1)+0.5) the vertical edge information of license plate area pixel, is strengthened with this; Wherein BYTE represents pixel gray-scale value, and abs is ABS function.
In described second step, utilize the car plate edge feature of regulation, pretreated image is scanned, reject the image border point not being inconsistent standard, and then obtain the license plate area of candidate.Its concrete steps are: the detection of marginal density.Due to the vertical edge abundant information of license plate area, and the horizontal edge density of background area is comparatively large, by the whole image of line scanning, for arbitrary pixel f (x, y) wherein, and using formula
Calculate marginal density threshold value T, wherein image_pixels is the number of pixels of whole image, and α is adjustable weights.Formulate the restrictive condition of each pixel, i.e. f (i, j) '=1, f (i, j) > f (i, j ± 1) & f (i, j) > T0, f (i, j) < T (3), when not meeting above-mentioned marginal density condition, this marginal point can be removed; The detection of edge length.After Image semantic classification, use three boundary scan method scan images, namely the length of the starting point at first time scanning marginal point range image top is designated as a1, the length of second time scanning distance image base terminal is designated as a2, third time is added first two length and is designated as a3, edge length scope [m, n] is set according to actual conditions, rejects corresponding marginal point with this.
In described 3rd step, adopt color dimension reduction method to process original image, obtain in second step with this and remain the corresponding color characteristic information of Edge Feature Points: utilize formula Y=(R+G)/2, RGB image is converted to YB two-dimensional space image; And calculate respective pixel point gray-scale value Gray=(Y+B)/2, wherein, R, G, B represent red, green, blue three kinds of Color Channel components respectively, and Y, Gray represent the Huang after process, ash two kinds of color components respectively;
As pixel value Y>187 and B>187 time, its pixel value is corrected as 255, is namely considered as white; As pixel value Y>153 and B>153 time, its pixel value is corrected as 0, is namely considered as black; In remaining Neutral colour, if Y<0.9*B, turn to step 3-5.
If B<0.9*Y, turn to step 3-6.Otherwise turn to step 3-7; Work as Y<0.8*B, be considered as blueness, otherwise Gray<187, be considered as blueness, be not so considered as grey; Work as B<0.8*Y, be considered as yellow, otherwise Gray>153, be considered as yellow, be not so considered as grey; Work as Gray<153, be considered as black, otherwise Gray>187, be considered as white, be not so considered as grey; Institute in this approach in scan image a little, and determines the color characteristic remaining marginal point in second step.
In described 4th step, the colouring information of combining image Edge Feature Points, finally accurately orients license plate area.By the process of the 3rd step, final whole license plate image is made up of blue, ash, yellow three kinds of colors.And for wrongly written or mispronounced character car plate of the blue end common at ordinary times, car plate font color is yellow after treatment, and car plate background color is blue, now font edge pixel point meets B<0.8Y|| (B>=0.8Y & & Gray>153) this condition.After processed for yellow end surplus car plate equally, car plate background color is yellow, and characters on license plate region is blue, now font edge pixel point meets Y<0.8*B|| (Y>=0.8B & & Gray<187).Car plate position can be locked further by the color characteristic in conjunction with this edge, thus positioning licence plate region.
Beneficial effect of the present invention is as follows:
(1) the present invention is based on gray image process, although used relevant colouring information feature, this algorithm does not use all information of RGB image, but utilize color dimension reduction method, only remain the colouring information for License Plate, finally change into into three value images, namely indigo plant, Huang and centre are grey, greatly reduce algorithm complex, therefore its travelling speed is higher, real-time.
(2) in rim detection, used multi-dimensional edge feature, the method namely having used several edge feature to combine accurately is located car plate.Improve relative to previous rim detection used and be mainly reflected in two aspects: in the third step, use color dimension reduction method to determine to remain the color characteristic information of marginal point, effectively can get rid of with this candidate license plate region utilizing conventional edge detection method to detect; This localization method avoids employing Binarization methods conventional in image segmentation processing method, thus avoids the uncertainty of binary-state threshold value.The improvement of these two aspects, decreases the calculated amount of algorithm of locating license plate of vehicle, adds computing velocity, also improves the location efficiency of car plate to a certain extent.
(3) be applied in Position System of automobile license plate location, have certain anti-interference to the unfavorable factor that the change of the conditions such as environment such as illumination, visual angle, Distance geometry background causes, also achieve good effect.Relative conventional method in addition, its computing velocity is faster, and real-time is better, and locating information is accurate.This method is applicable in Vehicle License Plate Recognition System, also can be widely used in the practices such as traffic intelligent management, computer vision.
Accompanying drawing explanation
Fig. 1 is FB(flow block) of the present invention.
Fig. 2 is gray level image.
Fig. 3 is the image after grey level enhancement.
The image of Fig. 4 after vertical edge detects.
Image after the scanning of Fig. 5 marginal density.
Image after the scanning of Fig. 6 edge length.
Fig. 7 carries out color dimension-reduction treatment to original image.
Image after Fig. 8 multi-dimensional edge detects.
Fig. 9 License Plate Segmentation location map.
Embodiment
Below in conjunction with accompanying drawing, the present invention is described further.
As shown in Figure 1, based on the localization method of multi-dimensional edge feature in Position System of automobile license plate location, first utilize relevant image pre-processing method to detect car plate marginal information, then utilize color dimension reduction method to original image process, be only preserved for the colouring information of License Plate.In conjunction with the multiple car plate marginal information feature after process, be applied in Position System of automobile license plate location, computing velocity is faster, and real-time is better, and locating information is accurate, and strong for the antijamming capability of the background of image.The method is applicable in the design ap-plication of Vehicle License Plate Recognition System, also can be widely used in the practices such as traffic intelligent management, computer vision
For better setting forth technical scheme of the present invention, the specific implementation step of its method is as follows:
Initial license plate image gray processing and grey level enhancement.In the present invention, we need first to the pre-service that initial pictures is correlated with, and is the image after gray level image and grey level enhancement as shown in Figures 2 and 3.And grey level enhancement is in order to license plate area outstanding in background, all pixel p=(x in scanning grey pictures
p, y
p), by enhancing function
calculate the pixel value of respective point, wherein I
pwith I'
pbe denoted as the pixel value before and after pixel p grey level enhancement respectively, W is arbitrary neighborhood of pixel points window,
for gray variance,
for average gray value.And
wherein T is used for representing and limits car plate marginal density threshold value, and the scope of T is (0,1), and be used for limiting marginal density approximate region scope, T is larger more on a large scale, and T is less more among a small circle.Beishu is used for adjusting
and the weights between Best, amplify when gray variance is near Best time.W is arbitrary neighborhood of pixel points window,
for gray variance, Best is average gray variance.By the contrast of above-mentioned two figure, just can better give prominence to the position of license plate area in whole image background.
Vertical edge operator is utilized to carry out rim detection.Major part license plate area has common marginal information feature, i.e. vertical edge abundant information.Vertical edge detection is carried out to the image after grey level enhancement, f (x is used to all pixel p, y)=BYTE (abs (f (x, y-1)-f (x, y+1)+0.5), the vertical edge information of license plate area pixel is strengthened with this, wherein, f (x, y) is the arbitrary pixel in image, BYTE represents pixel gray-scale value, and abs is ABS function.As shown in Figure 4, the vertical edge information of car plate is very abundant, but still has more similar noise region.
The detection of image border density.Due to the vertical edge abundant information of license plate area, and the horizontal edge density of background area is comparatively large, as shown in Figure 5, by the whole image of line scanning, for arbitrary pixel f (x, y) wherein, and using formula
Calculate marginal density threshold value T, wherein image_pixels is the number of pixels of whole image, and α is adjustable weights.Formulate the restrictive condition of each pixel, i.e. f (i, j) '=1, f (i, j) > f (i, j ± 1) & f (i, j) > T0, f (i, j) < T (3), when not meeting above-mentioned marginal density condition, this marginal point can be removed.
The detection of image border length.After Image semantic classification, for each pixel, use three boundary scan method scan images, namely the length of the starting point at first time scanning marginal point range image top is designated as a1, the length of second time scanning distance image base terminal is designated as a2, third time is added first two length and is designated as a3, according to actual conditions, edge length scope [m is set, n], m represents the minimum value of edge length, n represents edge length maximal value, be correlated with according to many factors such as concrete photo resolution to be analyzed, shooting focal length, angles in this region.Be set to m=5 herein, n=20.Corresponding marginal point is rejected with this.Fig. 6 can clearly find out, the marginal point of some ineligible edge length is all disallowable.
Utilize color dimension reduction method, obtain the color characteristic information of marginal point.The three-valued algorithm of image is by formula Y=(R+G)/2, RGB image space is mapped to YB space, again YB space is liked that you are simplified to the image of black, blue, grey, yellow, white five kinds of values, according to actual conditions, for the feature of Chinese car plate, blue board wrongly written or mispronounced character can be considered as yellow, yellow card surplus can also be considered as blueness, merged by black indigo plant, white yellow merging, be finally reduced to indigo plant, ash, yellow three value images.Algorithm thus, can determine license plate area character edge color characteristic information further.As shown in Figure 7, now font edge pixel point meets restrictive condition B<0.8Y|| (B>=0.8Y & & Gray>153).
In conjunction with three kinds of edge feature locating segmentation license plate areas.Previous step utilizes color dimension-reduction algorithm to obtain the color characteristic information of license plate area character edge point, by the candidate license plate region of screening in conjunction with second step two kinds of edge features, can accurately orient car plate position further.As shown in Figure 8, combine three kinds of edge features, the remaining marginal point of final rejecting is the marginal point of license plate area.Fig. 9 is the license plate area after segmentation.
Experimental situation of the present invention is: AMD athlon (tm) 64X2Dual core Processor 4800+2.40GHz, 2.00GB internal memory, the Realization of Simulation on MFC platform, obtain series of experiments result, wherein: in Fig. 3, optimum configurations is T=0.5, limits marginal density approximate region with this, amplify when gray variance is near Best time, and enlargement factor is controlled by Beishu; In Fig. 5, T0=5 being set, T2=15, when not meeting above-mentioned marginal density condition, this marginal point can being removed; Applied to color dimension-reduction algorithm in Fig. 7, although be based on coloured image License Plate, by dimensionality reduction, what finally obtain is three-valued image, and whole complex disposal process degree is lower, and working time is 15ms; In fig .9, by finally accurately orienting license plate area in conjunction with three kinds of edge features, 20ms consuming time altogether.
Although be described the specific embodiment of the present invention by reference to the accompanying drawings, in the claim limited range of this patent, the various amendment that those skilled in the art do not need creative work to make or distortion are still by the protection of this patent.
Claims (6)
1. based on a license plate locating method for multi-dimensional edge feature, it is characterized in that, its concrete steps are as follows:
The first step, with the former license plate image of image pre-processing method process, makes the vertical edge information displaying of former license plate image out;
Second step, utilizes the car plate edge feature of regulation, scans pretreated image, reject and be not inconsistent the image border point of standard, thus the partial noise in filtering image; In the image obtained, similar license plate area is as the candidate license plate region of the 4th step;
3rd step, adopts color dimension reduction method to process initial pictures, draws in second step and remain the corresponding color characteristic information of Edge Feature Points;
4th step, the colouring information of combining image Edge Feature Points, finally accurately orients license plate area.
2. as claimed in claim 1 based on the license plate locating method of multi-dimensional edge feature, it is characterized in that, in the described first step, during license plate image former in image pre-processing method process, first pre-service is carried out to image, obtain the image-region of marginal information; Its concrete steps are:
1-1) RGB image is converted to gray level image, grey level enhancement is carried out to the gray level image obtained;
1-2) to all pixel p=(x in gray level image
p, y
p) in its neighborhood window W, ask gray variance respectively
average gray value
and average gray variance Best;
1-3) by enhancing function
calculate the pixel value of respective point, wherein I
pwith I'
pbe denoted as the pixel value before and after pixel p grey level enhancement respectively, W is arbitrary neighborhood of pixel points window,
for gray variance,
for average gray value;
be gray variance near Best time amplify.
Vertical edge operator 1-4) is utilized to carry out rim detection to gray level image, vertical edge detection is carried out to the image after grey level enhancement, to all pixel p using formula f (x, y)=BYTE (abs (f (x, y-1)-f (x, y+1)+0.5) the vertical edge information of license plate area pixel, is strengthened; Wherein, f (x, y) is the arbitrary pixel in image, and BYTE represents pixel gray-scale value, and abs is ABS function.
3., as claimed in claim 2 based on the license plate locating method of multi-dimensional edge feature, it is characterized in that, described step 1-3) in, should
computing formula as follows:
4., as claimed in claim 1 based on the license plate locating method of multi-dimensional edge feature, it is characterized in that, in described second step, by the concrete steps of the former license plate image of image pre-processing method process be:
(2-1) detection of former license plate image marginal density; By the whole image of line scanning, for arbitrary pixel f (x, y) wherein, use computing formula:
Calculate marginal density threshold value T, wherein, image_pixels is the number of pixels of whole image, and α is adjustable weights; Formulate the marginal density restrictive condition of each pixel, i.e. f (i, j) '=1, f (i, j) > f (i, j ± 1) & f (i, j) > T0, f (i, j) < T (3), when not meeting above-mentioned marginal density restrictive condition, this marginal point can be removed;
(2-2) detection of edge length; After Image semantic classification, use three boundary scan method scan images, namely the length of the starting point at first time scanning marginal point range image top is designated as a1, and the length of second time scanning distance image base terminal is designated as a2, and third time is added first two length and is designated as a3, according to actual conditions, edge length scope [m is set, n], reject corresponding marginal point with this, wherein, m represents the minimum value of edge length, and n represents edge length maximal value.
5. as claimed in claim 1 based on the license plate locating method of multi-dimensional edge feature, it is characterized in that, in described 3rd step, the concrete steps adopting color dimension reduction method to process initial pictures are as follows:
3-1) utilize formula Y=(R+G)/2, RGB image is converted to YB two-dimensional space image, and calculate respective pixel point gray-scale value Gray=(Y+B)/2; Wherein, R, G, B represent red, green, blue three kinds of Color Channel components respectively, and Y, Gray represent the Huang after process, ash two kinds of color components respectively;
3-2) as pixel value Y>187 and B>187 time, its pixel value is corrected as 255, is namely considered as white;
3-3) as pixel value Y>153 and B>153 time, its pixel value is corrected as 0, is namely considered as black;
3-4) in remaining Neutral colour, if Y<0.9*B, turn to step 3-5); If B<0.9*Y, turn to step 3-6); Otherwise turn to step 3-7);
3-5) work as Y<0.8*B, be considered as blueness, otherwise Gray<187, be considered as blueness, all the other are considered as grey;
3-6) work as B<0.8*Y, be considered as yellow, otherwise Gray>153, be considered as yellow, all the other are considered as grey;
3-7) work as Gray<153, be considered as black, otherwise Gray>187, be considered as white, all the other are considered as grey;
Institute 3-8) in this approach in scan image a little, and determines the color characteristic remaining marginal point in second step.
6., as claimed in claim 1 based on the license plate locating method of multi-dimensional edge feature, it is characterized in that, in described 4th step, the colouring information of combining image Edge Feature Points, finally accurately orients license plate area;
By the process of the 3rd step, final whole license plate image is made up of blue, ash, yellow three kinds of colors;
When former car plate is blue end wrongly written or mispronounced character car plate, car plate font color is yellow after treatment;
When former car plate background color is blue, this former car plate font edge pixel point meets B<0.8Y|| (B>=0.8Y & & Gray>153) this condition;
When former car plate background color is black, after process, car plate background color is yellow, characters on license plate region is blue, and now font edge pixel point meets Y<0.8*B|| (Y>=0.8B & & Gray<187);
Car plate position can be locked further by the color characteristic in conjunction with this edge, thus positioning licence plate region.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410838603.9A CN104537350A (en) | 2014-12-30 | 2014-12-30 | License plate positioning method based on multi-dimensional edge feature |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410838603.9A CN104537350A (en) | 2014-12-30 | 2014-12-30 | License plate positioning method based on multi-dimensional edge feature |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104537350A true CN104537350A (en) | 2015-04-22 |
Family
ID=52852870
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410838603.9A Pending CN104537350A (en) | 2014-12-30 | 2014-12-30 | License plate positioning method based on multi-dimensional edge feature |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104537350A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106650553A (en) * | 2015-10-30 | 2017-05-10 | 比亚迪股份有限公司 | License plate recognition method and system |
| CN106874882A (en) * | 2017-02-24 | 2017-06-20 | 广东诚泰交通科技发展有限公司 | Pavement disease edge detection method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100246946A1 (en) * | 2009-03-27 | 2010-09-30 | Fuji Xerox Co., Ltd. | Image processing apparatus, image processing method and computer readable medium |
| CN103985135A (en) * | 2014-06-07 | 2014-08-13 | 山西中创伟业科技有限公司 | License plate location method based on difference edge images |
-
2014
- 2014-12-30 CN CN201410838603.9A patent/CN104537350A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100246946A1 (en) * | 2009-03-27 | 2010-09-30 | Fuji Xerox Co., Ltd. | Image processing apparatus, image processing method and computer readable medium |
| CN103985135A (en) * | 2014-06-07 | 2014-08-13 | 山西中创伟业科技有限公司 | License plate location method based on difference edge images |
Non-Patent Citations (2)
| Title |
|---|
| 余磊: "车牌字符快速识别关键算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 张浩鹏: "基于灰度方差和边缘密度的车牌定位算法", 《仪器仪表学报》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106650553A (en) * | 2015-10-30 | 2017-05-10 | 比亚迪股份有限公司 | License plate recognition method and system |
| CN106874882A (en) * | 2017-02-24 | 2017-06-20 | 广东诚泰交通科技发展有限公司 | Pavement disease edge detection method and device |
| CN106874882B (en) * | 2017-02-24 | 2020-02-18 | 广东诚泰交通科技发展有限公司 | Pavement disease edge detection method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103116751B (en) | A kind of Method of Automatic Recognition for Character of Lcecse Plate | |
| CN103077384B (en) | A kind of method and system of vehicle-logo location identification | |
| CN107891808B (en) | Driving reminding method and device and vehicle | |
| CN104751142A (en) | Natural scene text detection algorithm based on stroke features | |
| Li et al. | Nighttime lane markings recognition based on Canny detection and Hough transform | |
| CN107122777A (en) | A kind of vehicle analysis system and analysis method based on video file | |
| CN104899554A (en) | Vehicle ranging method based on monocular vision | |
| CN103927741A (en) | SAR image synthesis method for enhancing target characteristics | |
| Chen et al. | Detection and Recognition of Traffic Signs Based on HSV Vision Model and Shape features. | |
| CN104732211A (en) | Traffic sign detecting method based on self-adaptation threshold value | |
| CN106815583A (en) | A kind of vehicle at night license plate locating method being combined based on MSER and SWT | |
| Maya et al. | Performance analysis of lane detection algorithm using partial Hough transform | |
| CN103034983A (en) | Defogging method based on anisotropic filtering | |
| CN103049788B (en) | Based on space number for the treatment of object detection system and the method for computer vision | |
| CN103279755A (en) | Vehicle bottom shadow characteristics-based rapid license plate positioning method | |
| CN105893960A (en) | Road traffic sign detecting method based on phase symmetry | |
| CN106127765A (en) | Image binaryzation system based on self-adapting window and smooth threshold method | |
| Ingole et al. | Characters feature based Indian vehicle license plate detection and recognition | |
| CN111401364A (en) | License plate positioning algorithm based on combination of color features and template matching | |
| Wen et al. | A rear-vehicle detection system for static images based on monocular vision | |
| CN104835142A (en) | Vehicle queuing length detection method based on texture features | |
| Gu et al. | Vision-based multi-scaled vehicle detection and distance relevant mix tracking for driver assistance system | |
| CN104537350A (en) | License plate positioning method based on multi-dimensional edge feature | |
| CN109800693B (en) | Night vehicle detection method based on color channel mixing characteristics | |
| Chen et al. | Algorithm design of lane departure warning system based on image processing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150422 |
|
| RJ01 | Rejection of invention patent application after publication |