CN102073854A - Color license plate positioning method - Google Patents
Color license plate positioning method Download PDFInfo
- Publication number
- CN102073854A CN102073854A CN 201110008537 CN201110008537A CN102073854A CN 102073854 A CN102073854 A CN 102073854A CN 201110008537 CN201110008537 CN 201110008537 CN 201110008537 A CN201110008537 A CN 201110008537A CN 102073854 A CN102073854 A CN 102073854A
- Authority
- CN
- China
- Prior art keywords
- license plate
- area
- width
- binary map
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention belongs to the field of automatic license plate recognition and application of intelligent video monitoring, and relates to a color license plate positioning method. The method comprises the following steps of: acquiring CIELab color space automobile images; normalizing the images; performing morphological expansion processing on four classes of binary images by using diamond structure elements; obtaining a blue-white edge binary image, and obtaining a yellow-black edge binary image; performing expansion processing on the two edge binary images by using horizontal strip structure elements, and marking each potential license plate region; and setting four constraint conditions, namely constraint of license plate region area, relative size constraint of width and height of a license plate and width and height of an image, width-to-height ratio constraint of the license plate and shape constraint of the license plate, recognizing each potential license plate region and extracting the license plate region. The invention has the advantages that: the method is low in calculation quantity and high in processing speed, and is not affected by the interference region and the illumination condition change.
Description
Affiliated technical field
The car plate that the invention belongs to intelligent video monitoring is discerned application automatically, relates in particular to direct license plate locating method at coloured image.
Background technology
License plate identification (Vehicle License Plate Recognition abbreviates LPR as) technology is the important component part of intelligent transportation system.It is widely used in field of video monitoring such as unmanned parking lot, non-parking charge and magnitude of traffic flow control.LPR system based on computer vision mainly forms [1] by three parts: car plate is located, characters on license plate is cut apart and character recognition.Wherein, the car plate location is a key issue of LPR system, and what it was realized is the position of judging that license plate area occurs in image.
Domestic relevant car plate The Location starts from the nineties in last century, roughly Jing Li three developing stage.Early stage research is primarily aimed at gray level image, utilizes the edge or the textural characteristics of car plate.These class methods realize simple, and processing speed is fast, and are insensitive to the illumination condition variation, exist the positioning performance of situations such as inclination and distortion good for car plate, and weak point is that flase drop is more.On last class methods basis, the researcher is used for the car plate location in conjunction with various mathematical tools, comprising: mathematical morphology, genetic algorithm etc.Improving positioning performance simultaneously, the calculated amount of these class methods also significantly increases, and processing speed descends.Along with the raising of computing power, Color Image Processing becomes possibility on software and hardware is realized, increasing research steering coloured image car plate location.As [2] such as Chen Bin a kind of number-plate number searching method based on colouring information has been proposed, Guo Jie etc. [3] have proposed a kind of method based on color and texture analysis, to being used for the car plate location, Liu Wanjun etc. [5] use the cell neural network model to be used to detect the colour edging of license plate area to Li Wenju etc. [4] with edge color.These class methods have been taken all factors into consideration the color and the edge feature of car plate, can reduce flase drop, and weak point is that the car plate color is subject to the illumination condition variable effect, and omission is more.
In sum, existing car plate location technology is mainly used the edge feature location license plate area of car plate, and colouring information only plays booster action usually.Trace it to its cause, we find to have and carry out in the RGB color space at the method for colored license plate image.In this color space, three Color Channels (R, G, B) are relevant related, and any solid color passage all can not be described the gray scale and the colouring information of image independently.The CIELab color model is a kind of color model that has International Commission on Illumination (CIE) to announce in 1976.The CIELab color model is made up of a luminance channel (L) and two Color Channels (a and b).Wherein, the L passage represent brightness from black (value minimum) to white (value maximum); The color that a passage is represented is again to bright pink (value for just) from bottle green (value for negative) to grey (value is 0); The color that the b passage is represented is again to yellow (value for just) from sapphirine (value for negative) to grey (value is 0).
Pertinent literature
[1]Christos?Nikolaos?E.Anagnostopoulos,Ioannis?E.Anagnostopoulos,Vassili?Loumos,Eleftherios?Kayafas,A?license?plate-recognition?algorithm?for?intelligenttransportation?system?applications,IEEE?Transactions?on?IntelligentTransportation?Systems?7(3)(2006).377-392.
[2] Chen Bin, You Zhisheng. number-plate number color extracting searching method [J]. computer utility, 2001,21 (4): 74-75.
[3] Guo Jie, Shi Pengfei. based on the license plate locating method [J] of color and texture analysis. Chinese image graphics journal .2002,7 (5): 472-476.
[4] Li Wenju, Liang Dequn, Zhang Qi, etc. based on the right new location method of vehicle license plate of edge color [J], Chinese journal of computers, 2004,27 (2): 204-208.
[5] Liu Wanjun, Jiang Qingling opens and rushes. based on the license plate locating method [J] of CNN color images edge detection. and robotization journal, 2009,35 (12): 1503-1512.
Summary of the invention
The objective of the invention is to overcome the deficiencies in the prior art, it is little to propose a kind of calculated amount, and processing speed is fast, is difficult for the license plate locating method of the influence of disturbed area territory and illumination condition variation.Technical scheme of the present invention is as follows:
A kind of colored license plate locating method comprises the following steps:
1) monitoring camera is taken colored automobile image;
2) carry out color notation conversion space, obtain CIELab color space automobile image;
3) L, a, the b component of this image are done normalized, make span adjust to [0,1] between, is designated as L ', a ', b ' component respectively and uses L ', a ', b ' three-component to calculate the binary map of indigo plant, Huang, white, black four kinds of colors respectively, be designated as BL, YL, WH, BK respectively;
4) select for use " rhombus " structural motif respectively four class binary map to be carried out the morphology expansion process first time;
5) the blue color binary map and the white colour binary map of the process expansion process first time are asked union, obtain indigo plant-white edge edge binary map;
6) the yellow color binary map and the black color binary map of the process expansion process first time are asked union, obtain Huang-black surround edge binary map;
7) determine to carry out the size of the horizontal band-like structural motif of expansion process according to image resolution ratio, two edge binary map are carried out the expansion process second time respectively, each potential license plate area of mark;
8) width of the constraint of setting license plate area area, car plate and height retrain with the width and the relative size highly of image, car plate the ratio of width to height retrains, four constraint conditions of car plate shape constraining, each potential license plate area is discerned, extracted license plate area.
As preferred implementation, in the step 3), carry out binary conversion treatment according to following formula, in the formula (m n) is the pixel coordinate:
Preferably, the wide and height of establishing colored automobile image is respectively W and H, Ω
iRepresent certain potential license plate area, subscript i represents the sequence number that this is regional, AREA
iBe Ω
iThe sum of interior pixel; Use TOP
i, BOT
i, LET
iAnd RGT
iRepresent Ω respectively
iThe border, upper and lower, left and right, they are corresponding to Ω
iInterior each pixel coordinate value (x
i, y
i) middle x
iMinimum value and maximal value, y
iMinimum value and maximal value; Definition Ω
iWidth be WID
i=| RGT
i-LET
i|, highly be HET
i=| BOT
i-TOP
i|; Definition Ω
iThe ratio of width to height be WHR
i=WID
i/ HET
iJustice Ω
iThe boundary rectangle (TOP that serves as reasons
i, LET
i) and (BOT
i, RGT
i) 2 as the determined rectangle of diagonal line, uses ER
iExpression; With Ω
iRectangularity be defined as RECT
i=AREA
i/ (WID
i* HET
i), i.e. Ω
iArea and its ER
iThe ratio of area is for Ω
i, when having only it to satisfy all constraint conditions, just be judged to license plate area; Otherwise, be removed as interference region:
Condition 1:AREA
i>T
i(W * H), T
1=0.001;
Condition 2:0.05<WID
i/ W<0.25 and 0.02<HET
i/ H<0.1;
Condition 3:T
2<WHR
i<T
3, choose T
2=2, T
3=6;
Condition 4:RECT
i>T
4, T
4∈ [0.6,0.8].
In the CIELab space, use the b component can distinguish blue easily and yellow (background color of corresponding car plate), and the L component can effectively be represented white and black (character color of corresponding car plate).The present invention is according to the characteristics of CIELab color space, and in conjunction with color, edge and the shape facility of license plate area, proposes a kind of colored license plate locating method based on the CIELab color space.The present invention is by analyzing color, edge and the shape facility of license plate area, in conjunction with the characteristics of CIELab color space, extracts particular color edge corresponding to colored car plate to information, realizes the quick location of potential license plate area; Next in conjunction with shape facility, remove interference region, accurately locate license plate area.Use this method can detect wrongly written or mispronounced character car plate of the blue end (being called for short " blue car plate ") and yellow end surplus car plate (being called for short " yellow car plate ") automatically.Calculated amount of the present invention is little, and processing speed is fast, is difficult for the influence that disturbed area territory and illumination condition change, and is suitable for practical application.
Description of drawings
The FB(flow block) of Fig. 1 localization method of the present invention.
Fig. 3 operator 2 synoptic diagram that expand.
Embodiment
VC++2005 under the embodiment of the invention employing Windows XP SP3 system is as emulation platform.Test used automobile image the toll station real scene shooting automobile image that provides with cooperation unit is provided, totally 385 width of cloth contain 418 complete car plates.Wherein, daytime image 286 width of cloth, nighttime image 132 width of cloth, blue car plate 284 width of cloth, yellow car plate 134 width of cloth.The image resolution ratio that monitoring camera is taken is 1024 * 768pixels.
Technical scheme of the present invention mainly comprises three parts: color space conversion, license plate area are just located and license plate area is accurately located.Concrete scheme is as follows:
One, color space conversion
Take the colored automobile image obtain normally with the jpeg format storage by video camera, the color space that is adopted is sRGB (standard Red Green Blue) normally.The standard of using CIELab1976 to provide is finished the conversion of sRGB color space to the CIELab color space
At first, sRGB is converted to the CIEXYZ space, as the formula (1)
In the formula, R
S, G
S, B
SThe value of certain pixel R passage, G passage and B passage in the difference presentation video.Use formula (2) arrives the CIELab space with the CIEXYZ space conversion
Wherein
X in the formula (3)
n, Y
nAnd Z
nIn the expression CIEXYZ color space with reference to the tristimulus values of white point.Subscript n is represented normalized numerical value, adopts the tristimulus values under the D65 standard sources usually, and X is arranged this moment
n=95.047, Y
n=100.00, Z
n=108.88.In the CIELab color space that formula (1)~formula (3) obtains, the value of L is between [0,100], and the value of a component and b component is all between [128,127].
The common car plate of China has two classes at present, promptly blue end wrongly written or mispronounced character car plate (compact car) and yellow end surplus car plate (in/large car).This two classes car plate all has tangible colour edging information.For blue car plate, its inside has abundant indigo plant-Bai colour edging; For yellow car plate, its inside has abundant Huang-black surround edge.According to the characteristics of CIELab color space, use L component and b component to extract two category features easily.
Two, license plate area is just located
L, a, b component to input color image are done normalized, and span is all adjusted between [0,1], are designated as L ', a ', b '.Use L ', a ', b ' three-component to calculate the binary map of indigo plant, Huang, white, black race's color respectively, be designated as BL, YL, WH, BK respectively, as follows
In order to obtain colour edging figure, use morphology " expansion " computing that four binary map are handled.Consider the requirement of orientation independent, selecting radius for use is that " rhombus " structural motifs of 3 pixels is used for dilation operation.The expansion results of four class binary map is designated as BL respectively
D, YL
D, WH
D, BK
D, subscript D represents expansion results.Calculate BL
DAnd WH
0Union obtain indigo plant-white edge edge binary map, be designated as BW; Calculate YL
DAnd BK
DUnion obtain Huang-black surround edge binary map, be designated as YB.
According to the regulation of national standard, the character that China's license board information is arranged by a row or two row of horizontal constitutes.Characters on license plate comprises Chinese character, letter and number.All kinds of characters have tangible level of contrast and vertical stroke, and therefore, license plate area exists the marginal information that abundant horizontal direction is arranged.Based on above-mentioned analysis, this algorithm uses the morphology expansion process to extract potential license plate area from each colour edging binary map.
The shape of the used structural motif of dilation operation is very important.The edge of considering horizontal direction in the license plate area is more intensive than vertical direction, when selecting the expansion texture primitive, should pay the utmost attention to the marginal point that is communicated with vicinity on the horizontal direction.The present invention selects for use m * n horizontal band-like structural motif to be used for dilation operation, and m and n represent the width and the height of structural motif respectively, and their value is determined by automobile image resolution.With W and H represent license plate image width and the height, unit is a pixel count.Make discovery from observation, the overwhelming majority can discern the width of car plate greater than 1/10 of W, and the distance between interior adjacent two characters of license plate area is less than 1/10 of car plate width simultaneously.So desirable m=odd (W/100), the wherein computing of odd (x) expression " getting immediate odd number " with x, the value of n can be set according to the value of m, desirable n=odd (m/4).In the binary map that obtains after the expansion process, mark each connected region in the expansion results, and it potential license plate area as each connected region correspondence
Three, license plate area is accurately located
Select for use 4 kinds of shape facilities to be used for license plate area and differentiate, remove interference region.They are: area, relative width and height, the ratio of width to height and rectangularity.Use Ω
iRepresent certain potential license plate area, subscript i represents the sequence number that this is regional.4 kinds of shape facilities are defined as follows:
Area: Ω
iArea be defined as Ω
iThe sum of interior pixel is used AREA
iExpression.
Border up and down: use TOP
i, BOT
i, LET
iAnd RGT
iRepresent Ω respectively
iThe border, upper and lower, left and right, they are corresponding to Ω
iInterior each pixel coordinate value (x
i, y
i) middle x
iMinimum value and maximal value, y
iMinimum value and maximal value.
Width and height: definition Ω
iWidth be WID
i=| RGT
i-LET
i|, highly be HET
i=| BOT
i-TOP
i|.
The ratio of width to height: definition Ω
iThe ratio of width to height be WHR
i=WID
i/ HET
i
Boundary rectangle: definition Ω
iThe boundary rectangle (TOP that serves as reasons
i, LET
i) and (BOT
i, RGT
i) 2 as the determined rectangle of diagonal line, uses ER
iExpression.
Rectangularity: Ω
iRectangularity be defined as RECT
i=AREA
i/ (WID
i* HET
i), i.e. Ω
iArea and its ER
iThe ratio of area.Rectangularity has reflected the shape of license plate area and the degree of approximation of rectangle.
Use 4 shape constraining conditions of above-mentioned characteristics specify.For Ω
i, when having only it to satisfy all constraint conditions, just be judged to license plate area; Otherwise, be removed as interference region.
Condition 1:AREA
i>T
1(W * H)
That condition 2 is described is Ω
iWidth and height and the constraint of the relative size of W and H.The setting of bound is with reference to the analysis in the condition 1.Compare with condition 1, the constraint of condition 2 is more strict.
Condition 3:T
2<WHR
i<T
3
What condition 3 was described is to Ω
iThe constraint of the ratio of width to height.Under the normal condition, car plate the ratio of width to height is about 3.Consider license plate area may exist cover, situation such as inclination and distortion of projection, should relax constraint to the ratio of width to height.The present invention chooses T
2=2, T
3=6.
Condition 4:RECT
i>T
4
Condition 4 is the constraints to the car plate shape.Ideally car plate is shaped as rectangle.Although be subjected to distortion of projection, cover or the influence of factor such as damaged, cause the RECT of detected license plate area
iLess than 1, but generally speaking, the shape of license plate area still approaches rectangle.For reducing omission, threshold value Th
4Value can not be too big, common desirable T
4∈ [0.6,0.8], the present invention gets T
4=0.6.
After above-mentioned license plate area discrimination process processing, the residue connected region is judged to candidate's car plate.Consider that car plate may exist and tilt or damaged situation, for obtaining complete license plate area, the algorithm of carrying extension process is carried out on the border of candidate's license plate area.Concrete way is, according to the ER of each candidate's license plate area
i, determine its border, upper and lower, left and right.For left and right border, respectively to extending out
Individual pixel; For upper and lower border, respectively to extending out
Individual pixel, wherein
The expression " under round " computing.Use ER
iThe expression border extends out the result, according to ER
iFrom input picture, extract license plate area.
Use recall ratio (Recall) and precision ratio (Precision) as the index of weighing the algorithm performance of carrying.Represent car plate sum, N with N
cThe car plate number that expression is correctly oriented, N
fThe car plate number that the expression location of mistake goes out, the car plate number of omission is N
m=N-N
cRecall ratio is defined as N
cRatio with N.Precision ratio is defined as N
cCar plate number (the N that arrives with actual detected
c+ N
f) ratio.Table 1 has provided the positioning result that adopts method of the present invention.
Table 1 experimental result
By experimental result as can be seen, the present invention is practicable.The recall ratio of blue car plate is higher than yellow car plate, and main cause is that the visual quality of yellow car plate is poorer than blue car plate usually; The precision ratio of yellow car plate is a little more than blue car plate, and main cause is that the situation of the similar blue license plate area of appearance (as blue vehicle body pattern) in the background is more more.
Claims (3)
1. a colored license plate locating method comprises the following steps:
1) monitoring camera is taken colored automobile image;
2) carry out color notation conversion space, obtain CIELab color space automobile image;
3) L, a, the b component of this image are done normalized, make span adjust to [0,1] between, is designated as L ', a ', b ' component respectively and uses L ', a ', b ' three-component to calculate binary map blue, yellow, white, black four kinds of colors, be designated as BL, YL, WH, BK respectively;
4) select for use the diamond structure primitive respectively four class binary map to be carried out the morphology expansion process first time;
5) the blue color binary map and the white colour binary map of the process expansion process first time are asked union, obtain indigo plant-white edge edge binary map;
6) the yellow color binary map and the black color binary map of the process expansion process first time are asked union, obtain Huang-black surround edge binary map;
7) determine to carry out the size of the horizontal band-like structural motif of expansion process according to image resolution ratio, two edge binary map are carried out the expansion process second time respectively, each potential license plate area of mark;
8) width of the constraint of setting license plate area area, car plate and height retrain with the width and the relative size highly of image, car plate the ratio of width to height retrains, four constraint conditions of car plate shape constraining, each potential license plate area is discerned, extracted license plate area.
3. colored license plate locating method according to claim 1 is characterized in that, the wide and height of establishing colored automobile image is respectively W and H, Ω
iRepresent certain potential license plate area, subscript i represents the sequence number that this is regional, AREA
iBe Ω
iThe sum of interior pixel; Use TOP
i, BOT
i, LET
iAnd RGT
iRepresent Ω respectively
iThe border, upper and lower, left and right, they are corresponding to Ω
iInterior each pixel coordinate value (x
i, y
i) middle x
iMinimum value and maximal value, y
iMinimum value and maximal value; Definition Ω
iWidth be WID
i=| RGT
i-LET
i|, highly be HET
i=| BOT
i-TOP
i|; Definition Ω
iThe ratio of width to height be WHR
i=WID
i/ HET
iJustice Ω
iThe boundary rectangle (TOP that serves as reasons
i, LET
i) and (BOT
i, RGT
i) 2 as the determined rectangle of diagonal line, uses ER
iExpression; With Ω
iRectangularity be defined as RECT
i=AREA
i/ (WID
i* HET
i), i.e. Ω
iArea and its ER
iThe ratio of area is for Ω
i, when having only it to satisfy all constraint conditions, just be judged to license plate area; Otherwise, be removed as interference region:
Condition 1:AREA
i>T
1(W * H), T
1=0.001;
Condition 2:0.05<WID
i/ W<0.25 and 0.02<HET
i/ H<0.1;
Condition 3:T
2<WHR
i<T
3, choose T
2=2, T
3=6;
Condition 4:RECT
i>T
4, T
4∈ [0.6,0.8].
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201110008537 CN102073854A (en) | 2011-01-14 | 2011-01-14 | Color license plate positioning method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201110008537 CN102073854A (en) | 2011-01-14 | 2011-01-14 | Color license plate positioning method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN102073854A true CN102073854A (en) | 2011-05-25 |
Family
ID=44032388
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 201110008537 Pending CN102073854A (en) | 2011-01-14 | 2011-01-14 | Color license plate positioning method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102073854A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102254152A (en) * | 2011-06-17 | 2011-11-23 | 东南大学 | License plate location method based on color change points and color density |
CN103400121A (en) * | 2013-08-06 | 2013-11-20 | 河海大学 | License plate locating method based on colorful binary image |
CN104134079A (en) * | 2014-07-31 | 2014-11-05 | 中国科学院自动化研究所 | Vehicle license plate recognition method based on extremal regions and extreme learning machine |
CN105760856A (en) * | 2016-03-18 | 2016-07-13 | 中山大学 | License plate positioning method and system based on undirected graph segmentation |
CN106845479A (en) * | 2017-01-12 | 2017-06-13 | 山东大学 | A kind of small size detection method of license plate based on contrastive colours rectangular characteristic |
CN108154149A (en) * | 2017-12-08 | 2018-06-12 | 济南中维世纪科技有限公司 | Licence plate recognition method based on deep learning network share |
CN108460357A (en) * | 2018-03-14 | 2018-08-28 | 北京市公安局警卫局 | A kind of windowing alarm detection system and method based on image recognition |
CN108573195A (en) * | 2017-03-10 | 2018-09-25 | 中国科学院声学研究所 | A kind of barrel-shaped road roadblock detection method based on color of image pattern match |
CN110796698A (en) * | 2019-11-07 | 2020-02-14 | 厦门市美亚柏科信息股份有限公司 | Vehicle weight removing method and device with maximum area and minimum length-width ratio |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101408942A (en) * | 2008-04-17 | 2009-04-15 | 浙江师范大学 | Method for locating license plate under a complicated background |
CN101866553A (en) * | 2009-04-15 | 2010-10-20 | 付丽娜 | Location method of plate number based on difference and color pair characteristic point |
-
2011
- 2011-01-14 CN CN 201110008537 patent/CN102073854A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101408942A (en) * | 2008-04-17 | 2009-04-15 | 浙江师范大学 | Method for locating license plate under a complicated background |
CN101866553A (en) * | 2009-04-15 | 2010-10-20 | 付丽娜 | Location method of plate number based on difference and color pair characteristic point |
Non-Patent Citations (2)
Title |
---|
《制造业自动化》 20101031 徐雷 等 基于lab空间和图像形态学的车牌定位算法 第11-13,29页 1-3 第32卷, 第10期 * |
《计算机应用研究》 20100831 陈昌涛 等 Lab空间色彩分割在快速车牌定位中的应用 第3191-3193页 1-3 第27卷, 第8期 * |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102254152A (en) * | 2011-06-17 | 2011-11-23 | 东南大学 | License plate location method based on color change points and color density |
CN103400121A (en) * | 2013-08-06 | 2013-11-20 | 河海大学 | License plate locating method based on colorful binary image |
CN104134079A (en) * | 2014-07-31 | 2014-11-05 | 中国科学院自动化研究所 | Vehicle license plate recognition method based on extremal regions and extreme learning machine |
CN104134079B (en) * | 2014-07-31 | 2017-06-16 | 中国科学院自动化研究所 | A kind of licence plate recognition method based on extremal region and extreme learning machine |
CN105760856A (en) * | 2016-03-18 | 2016-07-13 | 中山大学 | License plate positioning method and system based on undirected graph segmentation |
CN106845479A (en) * | 2017-01-12 | 2017-06-13 | 山东大学 | A kind of small size detection method of license plate based on contrastive colours rectangular characteristic |
CN106845479B (en) * | 2017-01-12 | 2020-03-27 | 山东大学 | Small-size license plate detection method based on color contrast rectangle features |
CN108573195A (en) * | 2017-03-10 | 2018-09-25 | 中国科学院声学研究所 | A kind of barrel-shaped road roadblock detection method based on color of image pattern match |
CN108154149A (en) * | 2017-12-08 | 2018-06-12 | 济南中维世纪科技有限公司 | Licence plate recognition method based on deep learning network share |
CN108154149B (en) * | 2017-12-08 | 2021-12-10 | 济南中维世纪科技有限公司 | License plate recognition method based on deep learning network sharing |
CN108460357A (en) * | 2018-03-14 | 2018-08-28 | 北京市公安局警卫局 | A kind of windowing alarm detection system and method based on image recognition |
CN110796698A (en) * | 2019-11-07 | 2020-02-14 | 厦门市美亚柏科信息股份有限公司 | Vehicle weight removing method and device with maximum area and minimum length-width ratio |
CN110796698B (en) * | 2019-11-07 | 2022-11-29 | 厦门市美亚柏科信息股份有限公司 | Vehicle weight removing method and device with maximum area and minimum length-width ratio |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102073854A (en) | Color license plate positioning method | |
CN108108761B (en) | Rapid traffic signal lamp detection method based on deep feature learning | |
CN103971128B (en) | A kind of traffic sign recognition method towards automatic driving car | |
CN102364496B (en) | Method and system for identifying automobile license plates automatically based on image analysis | |
CN109284669A (en) | Pedestrian detection method based on Mask RCNN | |
CN103345766B (en) | A kind of signal lamp recognition methods and device | |
CN103761529B (en) | A kind of naked light detection method and system based on multicolour model and rectangular characteristic | |
CN103413147B (en) | A kind of licence plate recognition method and system | |
CN105160691A (en) | Color histogram based vehicle body color identification method | |
CN103824081B (en) | Method for detecting rapid robustness traffic signs on outdoor bad illumination condition | |
CN104809461B (en) | The licence plate recognition method and system of binding sequence image super-resolution rebuilding | |
CN102880863B (en) | Method for positioning license number and face of driver on basis of deformable part model | |
CN107016362B (en) | Vehicle weight recognition method and system based on vehicle front windshield pasted mark | |
CN104966066A (en) | Traffic block port monitoring oriented in-car human face detection method and system | |
CN109086687A (en) | The traffic sign recognition method of HOG-MBLBP fusion feature based on PCA dimensionality reduction | |
CN102722733A (en) | Identification method and device of license plate types | |
CN102737221B (en) | Method and apparatus for vehicle color identification | |
CN102819728A (en) | Traffic sign detection method based on classification template matching | |
CN103530608A (en) | Vehicle type judgment method and vehicle type judgment device | |
CN102999753A (en) | License plate locating method | |
CN104217440B (en) | A kind of method extracting built-up areas from remote sensing images | |
CN103425989A (en) | Vehicle color identification method based on significance analysis | |
CN104143077B (en) | Pedestrian target search method and system based on image | |
CN103914849A (en) | Method for detecting red date image | |
CN106815587A (en) | Image processing method and device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20110525 |