CN103927512B - vehicle identification method - Google Patents
vehicle identification method Download PDFInfo
- Publication number
- CN103927512B CN103927512B CN201410087512.6A CN201410087512A CN103927512B CN 103927512 B CN103927512 B CN 103927512B CN 201410087512 A CN201410087512 A CN 201410087512A CN 103927512 B CN103927512 B CN 103927512B
- Authority
- CN
- China
- Prior art keywords
- fit
- quadrangle
- horz
- calculating
- lines
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000012216 screening Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 7
- 230000001186 cumulative effect Effects 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 6
- 230000003044 adaptive effect Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 4
- 230000004931 aggregating effect Effects 0.000 claims description 3
- 230000002146 bilateral effect Effects 0.000 claims description 3
- 238000009499 grossing Methods 0.000 claims description 3
- 238000007689 inspection Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 6
- 230000011218 segmentation Effects 0.000 description 8
- 238000012544 monitoring process Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 1
- 230000003631 expected effect Effects 0.000 description 1
- 238000005206 flow analysis Methods 0.000 description 1
- 238000007429 general method Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011179 visual inspection Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The invention provides a reliable vehicle identification method. According to the vehicle identification method, the edge of a vehicle front face image is extracted through an improved Canny operator so that the outline of windshield on the front portion of a vehicle can be rapidly identified, the vehicle is identified by identifying the image characteristics within the range of the windshield, and the vehicle identification effect is improved.
Description
Technical Field
The present invention relates to a vehicle recognition method, and more particularly, to a method for recognizing a vehicle by recognizing a windshield contour at a front portion of the vehicle.
Background
At present, in the fields of urban traffic, municipal administration, public security and the like, the application of monitoring systems such as electronic eyes and the like is more and more popular. After the data such as videos and images acquired by the systems are analyzed and processed by a machine vision method, more comprehensive and deep information can be provided, and further scientific management and decision making of relevant management departments can be facilitated. For example, at key positions of urban main road intersections and highways, vehicle license plate recognition technology is adopted to help intelligent processing such as vehicle tracking, flow analysis and the like, and great success is achieved. However, when the requirements on social public safety are higher and higher, the tracking and identification of specific criminal suspect vehicles put higher requirements on image analysis technology: since the license plate of the suspected vehicle can be easily replaced, it is necessary to identify more characteristic information that is difficult to replace in the vehicle image.
Therefore, it is necessary to design a reliable vehicle identification method.
Disclosure of Invention
In view of the drawbacks of the prior art, the present invention provides a reliable vehicle identification method, which is implemented by identifying a windshield contour at a front portion of a vehicle, and comprises the following steps:
carrying out preprocessing of edge-preserving denoising on the image by using a bilateral filtering mode;
solving an edge graph by using a Canny operator of a regional adaptive threshold;
searching horizontal lines and vertical lines which are approximately horizontal in the edge graph, and aggregating and screening the lines;
performing straight line fitting on the transverse line, screening out the transverse line with high fitting degree, and calculating a main horizontal inclination angle;
screening and straight line fitting are carried out on the vertical lines;
according to the straight line fitting parameters, two horizontal lines are selected from the horizontal line set, and are combined with one line selected from the left vertical line set and the right vertical line set to form a quadrangle, and the quadrangle sets are formed by different combinations;
calculating the shape similarity and edge goodness of fit of each quadrangle in the quadrangle set and the vehicle windshield sample parameters, and calculating the weight of the quadrangle according to the shape similarity and edge goodness of fit;
selecting a quadrangle with the highest weight, taking four sides of the fitted quadrangle as a basic outline of the windshield, searching pixel points with higher gradient along the directions of the four sides of the quadrangle to complete and seal the outline of the windshield, and then performing curve filtering smoothing to obtain a final outline.
And extracting annual inspection marks and other obvious marks in the outline as image features to perform vehicle identification.
Preferably, the vehicle identification method comprises the following steps in the process of solving the edge map by using a Canny operator of the regional adaptive threshold value:
1) dividing the image into 16 sub-areas of 4 x 4;
2) in each subarea, calculating a cumulative gray histogram in the area;
wherein i is the ith gray level and ranges from 0 to 255, and I (g) is the number of pixels with the gray level of g in the image;
3) selecting a high threshold Th from a cumulative grayscale histogramhAnd a low threshold Thl；
Thl=0.4*Thh
4) And (3) operating a Canny operator according to the region threshold, when the operator is processed to a certain sub-region, processing by adopting the high and low thresholds in the region, and processing by adopting the threshold mean value of the adjacent sub-regions in the boundary neighborhood of the cross-region.
4. Preferably, the vehicle identification method is used for calculating the main horizontal inclination angle according to the horizontal line set Horz _ Lines in the process of performing straight line fitting on the horizontal Lines, screening out the horizontal Lines with high fitting degree and calculating the main horizontal inclination angle;
and voting to determine the main horizontal inclination Angle Horz _ Angle according to the horizontal Lines meeting the conditions in the Horz _ Lines set. The voting process is as follows:
1) determining an angle range [ Ang _ Min, Ang _ Max ] according to the k value range of each fitting straight line;
2) will [ Ang _ Min, Ang _ Max]Equally divided into 20 bins, each bin having an angular span of Angbin. Since the transverse line is screened to be between +/-10 degrees, AngbinLess than or equal to 1 degree;
3) for each horizontal line, the angle Ang is obtained according to the k value, and the ith bin center angle Ang closest to the angle is calculated according to the angleiThe ith bin and two bins adjacent to i are voted. The vote value votes are calculated as follows:
wherein,
4) for each bin, after the vote values of all horizontal lines are accumulated, the Horz _ Angle is determined as:
preferably, the vehicle identification method calculates the shape Similarity and edge goodness of fit of each quadrangle Q and the windshield sample parameter in the quadrangle set, and calculates the weight Priority of the quadrangle according to the shape Similarity and edge goodness of fit;
the calculation process is as follows:
1) selecting a plurality of vehicle samples of different types including cars, minivans, trucks, SUVs and the like, and manually selecting four points pt at the windshield of the vehicle1，pt2，pt3，pt4Based on this, the average upper/lower edge ratio t2b and the standard deviation σ are calculatedt2b(ii) a High to bottom ratio h2b and standard deviation σh2b(ii) a Mean base angle θ and standard deviation σθ；
2) For each quadrangle Q in the quadrangle setiThe parameters shown in FIG. 1 are calculated, and four characteristic values f are calculated1，f2，f3，f4：
f1＝(lent/lenb-t2b)2/σt2b 2
f2＝((h1+h2)/lenb/2-h2b)2/σh2b 2
f3＝(α-θ)2/σθ 2+(β-θ)2/σθ 2
f4＝(α-β)2/σθ 2
3) According to four characteristic values f1，f2，f3，f4Calculating the Similarity:
the edge goodness of fit Fitness reflects the four edges of Q and the four lines horz participating in the Q fittingt，horzbThe fit of left and right at the spatial position of the image is calculated as follows:
4) to horizontal line horztAll of them satisfy the condition pt1.x<=x<=pt4X set of points p (x)i,yi)|i=1,2,...NtAccording to horztLinear fitting parameter k oftAnd btCalculating the parameter fitt：
In the formula,
5) all the conditions pt satisfied on the left vertical line left1.y<=y<=pt2Y set of points p (x)i,yi)|i=1,2,...NlAccording to the straight line fitting parameter k of leftlAnd blCalculating the parameter fitl：
6) Horz was calculated in a similar mannerbAnd parameter fit of rightbAnd fitr；
7) According to fitt、fitl、fitbAnd fitrComputing Fitness:
Fitness=(fitt+fitl+fitb+fitr)/4
finally, Priority is calculated according to Similarity and Fitness:
Priority=0.7*Similarity+0.3*Fitness
compared with the prior art, the invention at least has the following technical effects: the vehicle identification method utilizes the image characteristics in the range of the vehicle windshield to identify the vehicle, and improves the vehicle identification effect.
Drawings
Fig. 1 shows a quadrilateral Q fitted with four lines, up, down, left, right, and the parameters used to calculate features in Q.
Detailed Description
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein.
The technical scheme adopted by the invention is as follows:
step 1: carrying out preprocessing of edge-preserving denoising on the image by using a bilateral filtering mode;
step 2: solving an edge graph by using a Canny operator of a regional adaptive threshold;
the Canny operator needs to set two thresholds of high and low, a general method adopts manual setting or global self-adaption determination, the method adopts a region self-adaption method to determine the threshold and operates the Canny operator, and the method comprises the following steps:
step 2-1: dividing the image into 16 sub-areas of 4 x 4;
step 2-2: in each subarea, calculating a cumulative gray histogram in the area;
wherein i is the ith gray level in the range of 0-255, and I (g) is the number of pixels with g gray level in the image, and a high threshold Th is selected according to the cumulative gray level histogramhAnd a low threshold Thl。
Thl=0.4*Thh
Step 2-3: and operating the Canny operator according to the region threshold value. When a certain sub-region is processed by a Canny operator, the high and low threshold values in the region are adopted for processing, and the threshold value mean value processing of the adjacent sub-region is adopted in the boundary neighborhood of the cross-region.
And step 3: searching horizontal lines and vertical lines which are approximately horizontal and vertical in the edge graph, and aggregating and screening the line set;
in the front face image of the vehicle, the vehicle windows, the engine cover, the vehicle roof, the skylight and the like have approximately horizontal transverse lines, and the vehicle side body and the left and right vertical beams of the vehicle windows have approximately vertical lines. The purpose of searching the horizontal and vertical lines is to determine candidate areas of the window position.
And 4, step 4: performing straight line fitting on each horizontal line in the horizontal line set Horz _ Lines, screening out the horizontal line with high fitting degree, and calculating a main horizontal inclination angle;
using a least squares method, a straight line y = kx + b is fitted from the point sequence of each transverse line, and the fitting degree is calculated as follows, where x iss,ysPoints in the horizontal line point sequence, len is the horizontal line length.
y＝k·xs+b
sigma＝∑(y-ys)2/len
Excluding the horizontal lines with excessive sigma, and voting the rest horizontal lines to determine the main horizontal inclination Angle Horz _ Angle. The voting process is as follows:
step 4-1: determining an angle range [ Ang _ Min, Ang _ Max ] according to the k value range of each fitting straight line;
step 4-2: will [ Ang _ Min, Ang _ Max]Equally divided into 20 bins, each bin having an angular span of Angbin. Since the transverse line is screened to be between +/-10 degrees, AngbinLess than or equal to 1 degree;
step 4-3: for each horizontal line, the angle Ang is obtained according to the k value, and the ith bin center angle Ang closest to the angle is calculated according to the angleiThe ith bin and two bins adjacent to i are voted. The vote value votes are calculated as follows:
wherein,
step 4-4: for each bin, after the vote values of all horizontal lines are accumulated, the Horz _ Angle is determined as:
and 5: screening and linearly fitting the left and right vertical line sets respectively in a similar manner to the step 4; considering that the vertical line is likely to be close to the vertical line to cause the too large fitting slope k to influence the calculation accuracy, reflecting x and y and then fitting to obtain a fitting straight line x = ky + b;
step 6: according to the straight line fitting parameters, two horizontal lines are selected from the horizontal line set, and the two horizontal lines and one line selected from the left vertical line set and the right vertical line set are combined into a quadrangle, and the quadrangle set is formed by different combinations. The method comprises the following steps:
step 6-1: selecting upper and lower transverse lines horz from transverse line settAnd horzbWherein the difference between the Angle of at least one of the bars and the main horizontal Angle Horz _ Angle is less than 5%. And horztAnd horzbThe degree of overlap in the horizontal x-direction cannot be less than half the length of the shorter of them, i.e.:
Overlap(horzt，horzb)>0.5*min(lenhorzt,lenhorzb)
step 6-2: selecting a left vertical line left and a right vertical line right from the left and right vertical line sets, wherein the overlapping degree of the left vertical line left and the right vertical line right in the y direction must be more than half of the length of the shorter vertical line;
step 6-3: calculate left, right and horz respectivelytAnd horzbFour intersection points pt ofi(x, y), i =1,2,3, 4. I.e. solving the following system of equations:
y＝kh·x+bh,h＝horzt,horzb
x＝kv·y+bv,v＝left,right
step 6-4: starting from the upper left corner, four intersection points are named as pt in a counterclockwise sequence1，pt2，pt3，pt4Forming a quadrilateral Q.
Step 6-5: and repeating the steps 6-1 to 6-4 to generate a quadrilateral set according to the horizontal and vertical line combination.
And 7: calculating the shape Similarity and edge goodness of fit of each quadrangle Q and the windshield sample parameter in the quadrangle set, and calculating the weight Priority of the quadrangle according to the shape Similarity and edge goodness of fit;
the shape Similarity degree Similarity reflects how similar the found quadrangle Q is to the shape of the windshield in the sample. The calculation process is as follows:
step 7-1: selecting 100 vehicle samples of different types including cars, vans, trucks, SUVs and the like, and manually selecting four points pt at the windshield of the vehicle1，pt2，pt3，pt4Based on this, the average upper-lower edge ratio t2b =0.7716 and the standard deviation σ is calculatedt2b= 0.0245; height to base ratio h2b =0.3570, standard deviation σh2b= 0.0346; base angle mean θ =72.12, standard deviation σθ=2.5373。
Step 7-2: for each quadrangle Q in the quadrangle setiThe parameters thereof as shown in FIG. 1 are calculated, and four eigenvalues f are calculated1，f2，f3，f4：
f1＝(lent/lenb-t2b)2/σt2b 2
f2＝((h1+h2)lenb/2-h2b)2/σh2b 2
f3＝(α-θ)2/σθ 2+(β-θ)2/σθ 2
f4＝(α-β)2/σθ 2
And 7-3: according to four characteristic values f1，f2，f3，f4Calculating the Similarity:
the edge goodness of fit Fitness reflects the four edges of Q and the four lines horz participating in the Q fittingt，horzbThe fit of left and right at the spatial position of the image is calculated as follows:
and 7-4: to horizontal line horztAll of them satisfy the condition pt1.x<=x<=pt4X set of points p (x)i,yi)|i=1,2,...NtAccording to horztLinear fitting parameter k oftAnd btCalculating the parameter fitt：
In the formula,
and 7-5: all the conditions pt satisfied on the left vertical line left1.y<=y<=pt2Y set of points p (x)i,yi)|i=1,2,...NlAccording to the straight line fitting parameter k of leftlAnd blCalculating the parameter fitl：
And 7-6: horz was calculated in a similar mannerbAnd parameter fit of rightbAnd fitr；
And 7-7: according to fitt、fitl、fitbAnd fitrComputing Fitness:
Fitness=(fitt+fitl+fitb+fitr)/4
finally, Priority is calculated according to Similarity and Fitness:
Priority=0.7*Similarity+0.3*Fitness
and 8: selecting the quadrangle Q with the highest weight Priority to fit four lines horz of Qt，horzbLeft and right are the basic outline of the windshield, the pixel points with higher gradient are searched along the four sides of Q to complete and seal the outline of the windshield, and then the pixel points are addedAnd filtering and smoothing the curve to obtain the final contour.
And step 9: and extracting annual inspection marks and other obvious marks in the outline as image features to perform vehicle identification.
The specific implementation case is as follows: and selecting part of samples from a vehicle front face image database collected by a traffic checkpoint monitoring system obtained from a certain public bureau, and obtaining front windshield sample parameters from the samples. And then, running a front windshield positioning program using the method of the invention to perform positioning test on other partial images in the database. Through tests, the segmentation efficiency and effect meet the expected requirements. And finally, operating the positioning program by a data center of a certain local police bureau, acquiring the monitoring image on line for segmentation and positioning, and storing the positioning result in an image form. After a period of time, the positioning result is detected and the vehicle identification program is integrated, so that the method provided by the invention can improve the identification effect and meet the actual requirements of users.
The implementation steps are as follows:
1. 500 front images of vehicles with resolution of 1024 × 768 of each of 100 types of 5 cars, minibuses, SUVs and buses are selected from the database.
2. And (3) selecting 100 images in total from the 500 images as sample images, and manually calibrating to obtain characteristic parameters: t2b =0.7716, σt2b=0.0245；h2b=0.3570，σh2b=0.0346；θ=72.12，σθ=2.5373。
3. The method of the invention is used for carrying out segmentation positioning on the windshield on the remaining 400 images, and the average segmentation time is 0.1367 seconds through statistics. The divided contours are marked on the original image and are marked as CaAnd storing.
4. The windshield outline of the 400 images was manually marked with a manual mark, denoted Cm。
5. With S (C)m) Is represented by CmArea of (d), S (C)a∩Cb) Is represented by CaAnd CmThe area of the overlapping portion, the ratio P (C) of the twoa,Cm)＝S(Ca∩Cm)S(Cm) To evaluate the segmentation effect, 400 images P (C) were obtaineda,Cm) Was found to have an average value of 93.76, as expected.
6. Running the program of the method on a data center server of a public security bureau, running for a period of time on line, randomly extracting 1000 images with segmentation results, and screening 78 images with poor segmentation results by adopting a visual inspection mode; the 78 images were then manually marked to calculate P (C)a,Cm) The average value was 83.21.
7. Marker information in the segmentation positioning range is extracted to serve as image features to assist vehicle identification, the identification rate is improved by 3 percent, and the expected effect is achieved.
And next, classifying and distinguishing the characteristic parameters aiming at different vehicle types, so that the positioning accuracy can be further improved.
Although the present invention has been described with reference to the preferred embodiments, it is not intended to be limited thereto, and variations and modifications may be made by those skilled in the art without departing from the spirit and scope of the present invention.
Claims (4)
1. A vehicle identification method characterized by: the vehicle identification method is realized by identifying the windshield outline at the front part of the vehicle, and comprises the following steps:
carrying out preprocessing of edge-preserving denoising on the image by using a bilateral filtering mode;
solving an edge graph by using a Canny operator of a regional adaptive threshold;
searching horizontal lines and vertical lines which are approximately horizontal in the edge graph, and aggregating and screening the lines;
performing straight line fitting on the transverse line, screening out the transverse line with high fitting degree, and calculating a main horizontal inclination angle; screening and straight line fitting are carried out on the vertical lines;
according to the straight line fitting parameters, two horizontal lines are selected from the horizontal line set, and are combined with one line selected from the left vertical line set and the right vertical line set to form a quadrangle, and the quadrangle sets are formed by different combinations;
calculating the shape similarity and edge goodness of fit of each quadrangle in the quadrangle set and the vehicle windshield sample parameters, and calculating the weight of the quadrangle according to the shape similarity and edge goodness of fit;
selecting a quadrangle with the highest weight, taking four sides of the fitted quadrangle as a basic outline of the windshield, searching pixel points with higher gradient along the directions of the four sides of the quadrangle to complete and seal the outline of the windshield, and then performing curve filtering smoothing to obtain a final outline;
and extracting annual inspection marks and other obvious marks in the outline as image features to perform vehicle identification.
2. The vehicle identification method according to claim 1, characterized in that: the vehicle identification method comprises the following steps in the process of solving an edge map by using a Canny operator of a regional adaptive threshold value:
1) dividing the image into 16 sub-areas of 4 x 4;
2) in each subarea, calculating a cumulative gray histogram in the area;
wherein i is the ith gray level and ranges from 0 to 255, and I (g) is the number of pixels with the gray level of g in the image;
3) selecting a high threshold Th from a cumulative grayscale histogramhAnd a low threshold Thl；
Thl＝0.4*Thh
4) And (3) operating a Canny operator according to the region threshold, when the operator is processed to a certain sub-region, processing by adopting the high and low thresholds in the region, and processing by adopting the threshold mean value of the adjacent sub-regions in the boundary neighborhood of the cross-region.
3. The vehicle identification method according to claim 1, characterized in that: the vehicle identification method comprises the steps of carrying out straight line fitting on a transverse line, screening out the transverse line with high fitting degree, and calculating a main horizontal inclination angle according to a transverse line set Horz _ Lines in the process of calculating the main horizontal inclination angle;
voting according to horizontal Lines meeting the conditions in the Horz _ Lines set to determine a main horizontal inclination Angle Horz _ Angle; the voting process is as follows:
1) determining an angle range [ Ang _ Min, Ang _ Max ] according to the k value range of each fitting straight line;
2) will [ Ang _ Min, Ang _ Max]Equally divided into 20 bins, each bin having an angular span of Angbin(ii) a Since the transverse line is screened to be between +/-10 degrees, AngbinLess than or equal to 10 degrees;
3) for each horizontal line, the angle Ang is obtained according to the k value, and the ith bin center angle Ang closest to the angle is calculated according to the angleiVoting for the ith bin and two bins adjacent to i; the vote value votes are calculated as follows:
wherein,
4) for each bin, after the vote values of all horizontal lines are accumulated, the Horz _ Angle is determined as:
4. the vehicle identification method according to claim 1, characterized in that: the vehicle identification method calculates the shape Similarity and edge matching degree Fitness of each quadrangle Q and the windshield sample parameter in the quadrangle set, and calculates the weight Priority of the quadrangle according to the shape Similarity and edge matching degree Fitness;
the calculation process is as follows:
1) selecting a plurality of vehicle samples of different types including cars, minivans, trucks, SUVs and the like, and manually selecting four points pt at the windshield of the vehicle1，pt2，pt3，pt4Based on this, the average upper/lower edge ratio t2b and the standard deviation σ are calculatedt2b(ii) a High to bottom ratio h2b and standard deviation σh2b(ii) a Mean base angle θ and standard deviation σθ；
2) For each quadrangle Q in the quadrangle setiThe parameters shown in FIG. 1 are calculated, and four characteristic values f are calculated1，f2，f3，f4：
f1＝(lent/lenb-t2b)2/σt2b 2
f4＝(α-β)2/σθ 2
3) According to four characteristic values f1，f2，f3，f4Calculating the Similarity:
the edge goodness of fit Fitness reflects the four edges of Q and the four lines horz participating in the Q fittingt，horzbThe fit of left and right at the spatial position of the image is calculated as follows:
4) to horizontal line horztAll of them satisfy the condition pt1.x<＝x<＝pt4X set of points p (x)i,yi)Ii＝1,2,...NtAccording to horztLinear fitting parameter k oftAnd btCalculating the parameter fitt：
In the formula,
5) all the left vertical lines left satisfy the condition pt1.y<＝y<＝pt2Y set of points p (x)i,yi)Ii＝1,2,...NlAccording to the straight line fitting parameter k of leftlAnd blCalculating the parameter fitl：
6) Technically similar method for calculating horzbAnd parameter fit of rightbAnd fitr：
7) According to fitt、fitl、fitbAnd fitrComputing Fitness:
Fitness＝(fittten fitlTen fitbTen fitr)/4
Finally, Priority is calculated according to Similarity and Fitness:
priority 0.7 Similarity 0.3 Fitness.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410087512.6A CN103927512B (en) | 2014-03-11 | 2014-03-11 | vehicle identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410087512.6A CN103927512B (en) | 2014-03-11 | 2014-03-11 | vehicle identification method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103927512A CN103927512A (en) | 2014-07-16 |
| CN103927512B true CN103927512B (en) | 2017-05-10 |
Family
ID=51145728
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410087512.6A Active CN103927512B (en) | 2014-03-11 | 2014-03-11 | vehicle identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103927512B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104167099A (en) * | 2014-08-13 | 2014-11-26 | 青岛盛嘉信息科技有限公司 | Electronic eye |
| CN104881630B (en) * | 2015-03-31 | 2018-12-04 | 浙江工商大学 | Vehicle identification method based on vehicle window segmentation and fuzzy feature |
| CN107016362B (en) * | 2017-03-30 | 2020-01-03 | 华中科技大学 | Vehicle weight recognition method and system based on vehicle front windshield pasted mark |
| CN107180231B (en) * | 2017-05-22 | 2018-03-23 | 宁夏金博乐食品科技有限公司 | Shared bicycle Weigh sensor system |
| CN107766802B (en) * | 2017-09-29 | 2020-04-28 | 广州大学 | Self-adaptive detection method for unbuckled safety belt of front-row driver and passenger of motor vehicle |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708691A (en) * | 2011-12-26 | 2012-10-03 | 南京信息工程大学 | False license plate identification method based on matching between license plate and automobile type |
| CN103207988A (en) * | 2013-03-06 | 2013-07-17 | 大唐移动通信设备有限公司 | Method and device for image identification |
| CN103324935A (en) * | 2013-06-27 | 2013-09-25 | 中山大学 | Method for vehicle positioning and region segmenting in image |
| CN103522982A (en) * | 2013-10-25 | 2014-01-22 | 公安部第三研究所 | Vehicle safety belt detection method and device based on image analysis |
-
2014
- 2014-03-11 CN CN201410087512.6A patent/CN103927512B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708691A (en) * | 2011-12-26 | 2012-10-03 | 南京信息工程大学 | False license plate identification method based on matching between license plate and automobile type |
| CN103207988A (en) * | 2013-03-06 | 2013-07-17 | 大唐移动通信设备有限公司 | Method and device for image identification |
| CN103324935A (en) * | 2013-06-27 | 2013-09-25 | 中山大学 | Method for vehicle positioning and region segmenting in image |
| CN103522982A (en) * | 2013-10-25 | 2014-01-22 | 公安部第三研究所 | Vehicle safety belt detection method and device based on image analysis |
Non-Patent Citations (1)
| Title |
|---|
| 一种自动车窗识别方法的设计与实现;骆玉荣;《计算机技术与应用进展》;20071230;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103927512A (en) | 2014-07-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Akagic et al. | Pavement crack detection using Otsu thresholding for image segmentation | |
| CN107729818B (en) | Multi-feature fusion vehicle re-identification method based on deep learning | |
| CN106203398B (en) | A kind of method, apparatus and equipment detecting lane boundary | |
| CN105373794B (en) | A kind of licence plate recognition method | |
| TWI381321B (en) | Method for image recombination of multiple images and identifying image and system for identifying image and outputting identification result | |
| DE102006012914B4 (en) | System and method for determining the distance to a preceding vehicle | |
| CN103927512B (en) | vehicle identification method | |
| CN107301405A (en) | Method for traffic sign detection under natural scene | |
| Ding et al. | An adaptive road ROI determination algorithm for lane detection | |
| CN107194393B (en) | Method and device for detecting temporary license plate | |
| CN109635733B (en) | Parking lot and vehicle target detection method based on visual saliency and queue correction | |
| CN102509098A (en) | Fisheye image vehicle identification method | |
| CN103268468A (en) | Automatic detection method for fastening of safety belts by front sitting persons on motor vehicle | |
| CN108256445B (en) | Lane line detection method and system | |
| CN104537348B (en) | Special vehicle recognition methods and its identifying system | |
| CN104299009A (en) | Plate number character recognition method based on multi-feature fusion | |
| CN110598550B (en) | Vehicle networking identification system based on multilayer information intelligent analysis | |
| CN105825169A (en) | Road-image-based pavement crack identification method | |
| CN110942640A (en) | Method for actively discovering suspect vehicle illegally engaged in network car booking passenger transportation | |
| CN104112141A (en) | Method for detecting lorry safety belt hanging state based on road monitoring equipment | |
| CN106650567A (en) | Seatbelt detection method and seatbelt detection device | |
| CN114299247A (en) | Rapid detection and problem troubleshooting method for road traffic sign lines | |
| Satzoda et al. | Robust extraction of lane markings using gradient angle histograms and directional signed edges | |
| CN104700068A (en) | SVM based detection method of safety belt of driver | |
| Ryu et al. | Feature-based pothole detection in two-dimensional images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20191226 Address after: 314413 No.2, Fengshou Avenue, Haining Economic and knitting industrial park, Jiaxing City, Zhejiang Province Patentee after: Zhejiang Haining Warp Knitting Industrial Park Development Co., Ltd Address before: Hangzhou City, Zhejiang province 310018 Xiasha Higher Education Park is 18 street. Patentee before: Zhejiang Gongshang University |