CN105740828A - Stop line detection method based on quick sign communication - Google Patents
Stop line detection method based on quick sign communication Download PDFInfo
- Publication number
- CN105740828A CN105740828A CN201610073505.XA CN201610073505A CN105740828A CN 105740828 A CN105740828 A CN 105740828A CN 201610073505 A CN201610073505 A CN 201610073505A CN 105740828 A CN105740828 A CN 105740828A
- Authority
- CN
- China
- Prior art keywords
- stop line
- image
- row
- connected domain
- information
- 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.)
- Granted
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
- G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
Abstract
The invention relates to a stop line detection method based on quick sign communication. The method comprises the following steps: collecting a pavement gray level picture, and extracting an interested area; preprocessing the extracted interested area to obtain a binary image; judging whether the image has a zebra crossing or not, detecting a stop line according to a relationship between the zebra crossing and the stop line if the image has the zebra crossing, and finishing stop line detection; otherwise, detecting whether the image has a pavement identifier or not; and if the image has the pavement identifier, detecting the stop line according to the relationship between the pavement identifier and the stop line to finish stop line detection. Through the statistics of connected domain information, auxiliary information, including the zebra crossing, the pavement identifier and the like, is used to jointly judge whether the stop line is in the presence or not and record the position of the stop line in the image.
Description
Technical field
The present invention relates to intelligent driving field, specifically a kind of stop line detection method based on Fast Labeling connection.
Background technology
Along with the development of society with science and technology, automobile plays more and more important role in daily life.While automobile brings convenience to people's lives, also the safety for people brings huge hidden danger, and particularly in road traffic crossing, this hidden danger is particularly acute.Automatic Pilot technology and unpiloted development, bring Gospel for solving this hidden danger.
Current automatic Pilot with unmanned in, frequently with the stop line detection technique based on Hough transformation.This technology is in the gray level image collected, and detects " straight line " in image with preconditioning technique, image border technology, Hough transform technique, and judges whether stop line by apriority knowledge such as the angles of straight line.Because Hough transformation algorithm is very consuming time, cause that algorithm is overall consuming time higher, it is impossible to meet the demand of present vehicle-mounted Real-time System.
In intelligent driving image domains, labelling method for communicating has extremely important effect as a basic skills and is widely applied.In the detection of stop line, no matter it is cutting area-of-interest or the judgement of stop line position, is required for labelling UNICOM function.And present labelling connecting function, connected region often can only be set to identical value, provide a connected domain matrix.And the statistical information such as the width of the number of point, the height of connected domain, connected domain in the single connected domain initial row in the picture that we need, termination row, initial row, end column, connected domain, all without providing, it is necessary to provided by statistics connected domain matrix.And again add up, not only consume the substantial amounts of time, be also easy to make the mistake.
Summary of the invention
For the deficiencies in the prior art, the present invention provides a kind of stop line detection method based on Fast Labeling connection, by adding up connected domain information, and utilize the auxiliary information such as zebra crossing, road surface identification symbol jointly to complete the presence or absence judgement of stop line, and record the position at stop line place in the picture, quickly provide in image, the statistical information of each connected region, be used for accelerating the detection speed of stop line.
The present invention be the technical scheme is that for achieving the above object
A kind of stop line detection method based on Fast Labeling connection, comprises the following steps:
Step 1: gather road surface gray scale picture, extracts area-of-interest；
Step 2: the area-of-interest extracted is carried out pretreatment and obtains binary map；
Step 3: judge whether there are zebra crossing in image, if it is present detect stop line according to the relation between zebra crossing and stop line, completes stop line detection；Otherwise whether detection image exists road surface identification symbol；
Step 4: if there is road surface identification symbol, then according to the relation detection stop line between road surface identification symbol and stop line, complete stop line detection.
The process of described extraction area-of-interest is, this trapezoid area, by row neighbour's differential technique, is converted into rectangular area, is area-of-interest by the trapezoidal region in a track；
Described row neighbour's differential technique is: by the row of random length, by neighbour's difference, be stretched as the row of specified width, which width.
Described preprocessing process is: region of interest area image carries out large scale mean filter process, obtains target image, then target image is carried out binary conversion treatment, obtain binary map.
Described judge that whether there are zebra crossing in image includes procedure below:
Step 1: extract the image the first row half-tone information characteristic vector as detection zebra crossing；
Step 2: by the characteristic vector input adboost neutral net of detection zebra crossing, it is judged that whether this characteristic vector is zebra crossing, if it is, there are zebra crossing in image, it is judged that terminate, and otherwise performs step 3；
Step 3: judge whether current line is image last column, if it is, extract next line half-tone information, as the characteristic vector of detection zebra crossing, performs step 2, is otherwise absent from zebra crossing in image, it is judged that terminate.
The described relation according between zebra crossing and stop line detects stop line and includes procedure below:
Step 1: added up by current line gray value, obtains the cumulative sum of current line, is designated as Sum_Zebra；
Step 2: image is gone cumulative, obtains cumulative sum array；
Step 3: find the row less than 0.8*Sum_Zebra in cumulative sum array, be designated as R1；
Step 4: from R1 row, finds the row more than 1.4*Sum_Zebra in cumulative sum array；
Step 5: if it is found, then behavior stop line is expert at, complete detection process；
Step 6: otherwise image is carried out labelling connection, and adds up connected domain information；
Step 7: if there is connected domain width more than the 2/3 of picture traverse, then the initial behavior stop line of connected domain is expert at, and completes detection process；Otherwise it is absent from stop line, completes detection process.
The described relation detection stop line accorded with between stop line according to road surface identification includes procedure below:
Step 1: extract road surface identification symbol characteristic vector；
Step 2: by the characteristic vector input adboost neutral net of detection road surface identification symbol, it is judged that whether this characteristic vector is road surface identification symbol, if, image then exists road surface identification symbol, perform step 3, be otherwise absent from road surface identification symbol, complete stop line detection process；
Step 3: intercept image the first row to the parts of images between identifier initial row, parts of images is carried out labelling connection, and adds up connected domain information；
Step 4: if there is connected domain width more than the 2/3 of picture traverse, then the initial behavior stop line of connected domain is expert at, and completes detection process；Otherwise it is absent from stop line, completes detection process.
Described labelling handshaking procedures is:
Step 1: gather gray level image；
Step 2: to the gray level image average-histogram method gathered, try to achieve binary-state threshold, obtain binary map；
Step 3: connected domain information single in statistics the first row；
Step 4: the connected domain information of current line is saved in information array；
Step 5: determine whether image last column, if it is, information array statistics terminates, performs step 7；Otherwise, connected domain information single in statistics next line, and perform step 6；
Step 6: judge whether current line connected domain information has overlapping with lastrow connected domain information, if it has, then update current information array, otherwise performs step 4；
Step 7: the information array according to statistics, the connected domain information existed in statistical picture.
Described average-histogram method is:
Step 1: the rectangular histogram of statistics gray level image, is designated as piA；The meansigma methods of the image intensity value asked is:
Wherein, piA is the rectangular histogram of storage image, and iH is the height of image, and iW is the width of image, and aver is the meansigma methods of image intensity value；
Step 2: utilize histogram calculation to be not more than the meansigma methods of gray value of meansigma methods aver of image intensity value:
In like manner calculate the meansigma methods of the gray value of the meansigma methods aver more than image intensity value:
Wherein, aver1 is the meansigma methods of the gray value of the meansigma methods aver being not more than image intensity value, and aver2 is the meansigma methods of the gray value of the meansigma methods aver more than image intensity value；
Whether < 5 sets up step 3: Rule of judgment | aver-(aver1+aver2)/2 |, if setting up, then the meansigma methods aver of image intensity value is exactly required binary-state threshold, otherwise aver=(aver1+aver2)/2, returns step 2.
The connected domain information existed in described statistical picture includes procedure below: retrieve the 4th identical row in information array；4th identical row is added up the cumulative sum of primary minima in the 4th identical row, deputy maximum and the 3rd；4th identical row is added up minimum row sequence and maximum row sequence in the 4th identical row；According to the row, column information counted, calculating the initial of connected domain is classified as primary minima in the 4th identical row, the termination of connected domain is classified as deputy maximum in the 4th identical row, in connected domain, the number of pixel is the cumulative sum of the 3rd in the 4th identical row, minimum row sequence in the 4th identical row of the initial behavior of connected domain, maximum row sequence in the 4th identical row of the termination behavior of connected domain.
Described information array is a columns is the array of 5, in storing one row, and the connected domain information of single connected region；In every a line of information array, first row being used for storing the starting pixels point place of the single connected domain of current line, second is used for storing the row terminating pixel place of the single connected domain of current line, 3rd number being used for storing the pixel of single connected domain, 4th mark value being used for storing this connected domain, the 5th the line order row being used for storing current line；The mark value of connected domain is an increasing sequence by 2.
The invention have the advantages that and advantage:
1. utilize stop line that column direction is insensitive, it is possible to area-of-interest carries out large scale row sampling, and then reduction processes the size of image, it is ensured that described method disclosure satisfy that the demand of system real time.Speed is 5-10 times of common Hough transformation algorithm；
2. the detection of stop line, assists with zebra crossing and road surface road surface identification symbol.Improve the accuracy of stop line detection, eliminate the interference to algorithm of the road surface complex road condition, enhance the robustness of program；
3. the present invention is while statistics connected domain, provides the information of single connected domain in image；
4. the present invention adds up the speed quickening of connected domain, and the labelling connection algorithm with single connected domain information is more than ten times that conventional tag connection realizes.
Accompanying drawing explanation
Fig. 1 is the method flow diagram of the present invention；
Fig. 2 is the stop line detection schematic diagram of the present invention；
Fig. 3 is the labelling connection method flow chart of the present invention；
Fig. 4 is the average-histogram method of the present invention；
Fig. 5 is the information array schematic diagram of the present invention.
Detailed description of the invention
Below in conjunction with drawings and Examples, the present invention is described in further detail.
The flow chart of the present invention is as shown in Figure 1.The present invention detects part and realizes being segmented into three parts.Part I, it may be judged whether have zebra crossing and road surface identification symbol.If Part II has zebra crossing, then associating zebra crossing determine whether stop line.Part III, if there being road surface identification to accord with, associating road surface identification symbol determines whether stop line.
Part I, it may be judged whether have zebra crossing and road surface road surface identification symbol.First, in collecting gray level image, area-of-interest, the position that namely stop line is likely to occur are selected.Judge the presence or absence of stop line together to combine zebra crossing, the present invention selects a trapezoid area as area-of-interest.In the process of actual treatment, trapezoid area is stretched as rectangular area by linear change, the method adopting linear difference, so far extract area-of-interest and be partially completed.Next image being carried out pretreatment, primary operational processes for image carries out statistics with histogram denoising point, it is therefore an objective to kill the brighter part of image and dark part, gets rid of the complex road surface impact on algorithm.
Present invention Adboost algorithm judges the presence or absence of zebra crossing.Adboost algorithm is divided into training and two parts of test.Image after pretreatment, every a line can regard an Adboost sample.Adboost network parameter is obtained by the training of large sample.At test phase, often extract a line and form characteristic vector, by characteristic vector to Adboost network, obtain whether this feature is zebra crossing.
If zebra crossing, then Part I terminates, and is recorded by current line, and zebra crossing exist mark and are set to 1.Road surface identification symbol feature is then extracted, it may be judged whether accord with for road surface identification if not zebra crossing.Stop line it not that road surface identification accords with then EP (end of program), image do not have stop line, because must occur in pairs with road surface identification symbol or zebra crossing, if there is road surface identification symbol, then Part I terminates, and road surface identification symbol flag bit is designated as 1, and records road surface identification symbol initial row in the picture.
When zebra crossing flag bit is 1, the Part II of starting algorithm.The problem that Part II mainly solves is, when there are zebra crossing, how to detect whether there is stop line.Fig. 2 is the stop line detection schematic diagram of the present invention.First image is gone cumulative, obtain cumulative sum array, from the row at zebra crossing place, find pixel value be expert at than zebra crossing and little a lot of row, generally take the half of the be expert at cumulative sum of zebra crossing.Blank parts between behavior zebra crossing and stop line.From blank parts, cumulative sum is looked for be expert at much larger row than zebra crossing, generally take that zebra crossing are expert at 1.7 times.Because stop line shows as complete white in gray level image, and zebra crossing are chequered with black and white, the gray value of the row at stop line place and be typically in more than 1.7 times of the be expert at cumulative sum of zebra crossing.If it is found, then may determine that stop line position in the picture, EP (end of program), record stop line position in the picture.Without finding qualified row, then by add up labelling communication information, look for and whether there is a connected domain, his width more than figure image width 2/3rds.If it is present may determine that stop line position in the picture, EP (end of program), record stop line position in the picture is the initial row of this connected region.If it does not exist, then this image is absent from stop line, EP (end of program).
When identifier flag bit is 1, the Part III of starting algorithm.The problem that Part III mainly solves is, when there is road marking and not having zebra crossing, how to detect whether there is stop line.First pass through the labelling communication information of statistical picture, look for and whether there is a connected domain, his width more than figure image width 2/3rds.If it is present may determine that stop line position in the picture, EP (end of program), record stop line position in the picture is the initial row of this connected region.If it does not exist, then this image is absent from stop line, EP (end of program).
It is illustrated in figure 3 the labelling connection flow chart of the present invention.In the gray level image that early stage gathers, first image is carried out binary conversion treatment., such as Da-Jin algorithm, local binarization method etc., there is time and the opposed situation of effect in traditional binaryzation means, time and effect are difficult to take into account.There is employed herein a kind of method of quick rectangular histogram-average binaryzation and image is carried out binary conversion treatment, the method is the overall binarization method of a kind of similar Da-Jin algorithm, but in hgher efficiency compared with Da-Jin algorithm, and result is more or less the same.
It is illustrated in figure 4 the average-histogram method of the present invention.First add up the rectangular histogram of gray level image, be designated as piA.Utilize the rectangular histogram of image, it is possible to the meansigma methods of the image intensity value asked is:
Wherein piA stores the rectangular histogram of image, and iH is the height of image, and iW is the width of image, and i belongs to (0,255)
Utilize the meansigma methods of the histogram calculation gray scale less than or equal to aver
In like manner calculate the meansigma methods of the gray scale more than aver
Whether < 5 sets up Rule of judgment | aver-(aver1+aver2)/2 |, if setting up, then aver is exactly required binary-state threshold, otherwise aver=(aver1+aver2)/2, repeat the above steps.
It is illustrated in figure 5 the information array schematic diagram of the present invention.
Utilize the threshold value tried to achieve that image is carried out binary conversion treatment and obtain binary map.Binary map is added up the communication information of each connected region of the first row, and the communication information of each connected region is stored in " information array ".In the first row, there are two connected domains.First connected domain is from the 1st pixel to the 3rd pixel.Second connected domain is from the 8th pixel to the 9th pixel.
" information array ", in storing one row, the communication information of single connected region, information array piY is a width is the array of 5.In every a line, first be used for storing the starting pixels point of the single connected domain of one's own profession, second for store the single connected domain of one's own profession terminate pixel, the 3rd for store the length of single connected domain, the 4th for storing the mark value of this connected domain, the 5th for storing the line number of one's own profession.The mark value of the connected domain of the 4th storage is an increasing sequence by 2.
After having added up the first row communication information, add up the communication information of the second row.If the communication information of the second row and the first row have lap.Then update information array.If there is no lap, then carry out next line.Second row has two connected domains, and the connected domain of first connected domain and the first row has handing-over.So the two connected domain to be merged into a connected domain, the mark value of the two connected domain is designated as the mark value of the second row connected domain.The connected domain with the first row of second second connected domain of row does not have the part superposed, and therefore need not change information array.
After completing the second line flag range statistics, carry out next line, the statistics of single connected domain.And check whether this connected domain superposes with the connected domain of lastrow.If there being superposition, need the 4th connected region mark value of renewal information array.By that analogy, update mark connection " information array ", completes the statistics of the connected region of entire image gradually.
After completing the statistics of full figure connected region, it is necessary to the information of single connected domain in statistical picture.Now utilize labelling communication information array, array is retrieved the 4th identical row.To the 4th identical row, adding up the value that in all row, first minimum value is maximum with second, complete the statistics single connected domain information initiateing row with end column, the initial row of this connected domain and termination row then can be added up by the 5th.By the 4th identical row, the 3rd bit value adds up, it is possible to count in this single connected region, promising 1 the number of point.The Gao Yukuan of this connected domain can be calculated by the initial row, column of this connected domain, complete the statistical work of single connected domain in image.
Claims (10)
1. the stop line detection method based on Fast Labeling connection, it is characterised in that: comprise the following steps:
Step 1: gather road surface gray scale picture, extracts area-of-interest；
Step 2: the area-of-interest extracted is carried out pretreatment and obtains binary map；
Step 3: judge whether there are zebra crossing in image, if it is present detect stop line according to the relation between zebra crossing and stop line, completes stop line detection；Otherwise whether detection image exists road surface identification symbol；
Step 4: if there is road surface identification symbol, then according to the relation detection stop line between road surface identification symbol and stop line, complete stop line detection.
2. the stop line detection method based on Fast Labeling connection according to claim 1, it is characterized in that: the process of described extraction area-of-interest is, the trapezoidal region in a track, by row neighbour's differential technique, this trapezoid area is converted into rectangular area, is area-of-interest；
Described row neighbour's differential technique is: by the row of random length, by neighbour's difference, be stretched as the row of specified width, which width.
3. the stop line detection method based on Fast Labeling connection according to claim 1, it is characterised in that: described preprocessing process is:
Region of interest area image is carried out large scale mean filter process, obtains target image, then target image is carried out binary conversion treatment, obtain binary map.
4. the stop line detection method based on Fast Labeling connection according to claim 1, it is characterised in that: described judge that whether there are zebra crossing in image includes procedure below:
Step 1: extract the image the first row half-tone information characteristic vector as detection zebra crossing；
Step 2: by the characteristic vector input adboost neutral net of detection zebra crossing, it is judged that whether this characteristic vector is zebra crossing, if it is, there are zebra crossing in image, it is judged that terminate, and otherwise performs step 3；
Step 3: judge whether current line is image last column, if it is, extract next line half-tone information, as the characteristic vector of detection zebra crossing, performs step 2, is otherwise absent from zebra crossing in image, it is judged that terminate.
5. the stop line detection method based on Fast Labeling connection according to claim 1, it is characterised in that: the described relation according between zebra crossing and stop line detects stop line and includes procedure below:
Step 1: added up by current line gray value, obtains the cumulative sum of current line, is designated as Sum_Zebra；
Step 2: image is gone cumulative, obtains cumulative sum array；
Step 3: find the row less than 0.8*Sum_Zebra in cumulative sum array, be designated as R1；
Step 4: from R1 row, finds the row more than 1.4*Sum_Zebra in cumulative sum array；
Step 5: if it is found, then behavior stop line is expert at, complete detection process；
Step 6: otherwise image is carried out labelling connection, and adds up connected domain information；
Step 7: if there is connected domain width more than the 2/3 of picture traverse, then the initial behavior stop line of connected domain is expert at, and completes detection process；Otherwise it is absent from stop line, completes detection process.
6. the stop line detection method based on Fast Labeling connection according to claim 1, it is characterised in that: the described relation detection stop line accorded with between stop line according to road surface identification includes procedure below:
Step 1: extract road surface identification symbol characteristic vector；
Step 2: by the characteristic vector input adboost neutral net of detection road surface identification symbol, it is judged that whether this characteristic vector is road surface identification symbol, if, image then exists road surface identification symbol, perform step 3, be otherwise absent from road surface identification symbol, complete stop line detection process；
Step 3: intercept image the first row to the parts of images between identifier initial row, parts of images is carried out labelling connection, and adds up connected domain information；
Step 4: if there is connected domain width more than the 2/3 of picture traverse, then the initial behavior stop line of connected domain is expert at, and completes detection process；Otherwise it is absent from stop line, completes detection process.
7. the stop line detection method based on Fast Labeling connection according to claim 5 or 6, it is characterised in that: described labelling handshaking procedures is:
Step 1: gather gray level image；
Step 2: to the gray level image average-histogram method gathered, try to achieve binary-state threshold, obtain binary map；
Step 3: connected domain information single in statistics the first row；
Step 4: the connected domain information of current line is saved in information array；
Step 5: determine whether image last column, if it is, information array statistics terminates, performs step 7；Otherwise, connected domain information single in statistics next line, and perform step 6；
Step 6: judge whether current line connected domain information has overlapping with lastrow connected domain information, if it has, then update current information array, otherwise performs step 4；
Step 7: the information array according to statistics, the connected domain information existed in statistical picture.
8. the stop line detection method based on Fast Labeling connection according to claim 7, it is characterised in that: described average-histogram method is:
Step 1: the rectangular histogram of statistics gray level image, is designated as piA；The meansigma methods of the image intensity value asked is:
Wherein, piA is the rectangular histogram of storage image, and iH is the height of image, and iW is the width of image, and aver is the meansigma methods of image intensity value；
Step 2: utilize histogram calculation to be not more than the meansigma methods of gray value of meansigma methods aver of image intensity value:
In like manner calculate the meansigma methods of the gray value of the meansigma methods aver more than image intensity value:
Wherein, aver1 is the meansigma methods of the gray value of the meansigma methods aver being not more than image intensity value, and aver2 is the meansigma methods of the gray value of the meansigma methods aver more than image intensity value；
Whether < 5 sets up step 3: Rule of judgment | aver-(aver1+aver2)/2 |, if setting up, then the meansigma methods aver of image intensity value is exactly required binary-state threshold, otherwise aver=(aver1+aver2)/2, returns step 2.
9. the stop line detection method based on Fast Labeling connection according to claim 7, it is characterised in that: the connected domain information existed in described statistical picture includes procedure below:
Information array is retrieved the 4th identical row；4th identical row is added up the cumulative sum of primary minima in the 4th identical row, deputy maximum and the 3rd；4th identical row is added up minimum row sequence and maximum row sequence in the 4th identical row；According to the row, column information counted, calculating the initial of connected domain is classified as primary minima in the 4th identical row, the termination of connected domain is classified as deputy maximum in the 4th identical row, in connected domain, the number of pixel is the cumulative sum of the 3rd in the 4th identical row, minimum row sequence in the 4th identical row of the initial behavior of connected domain, maximum row sequence in the 4th identical row of the termination behavior of connected domain.
10. the stop line detection method based on Fast Labeling connection according to claim 7 or 9, it is characterised in that: described information array is a columns is the array of 5, in storing one row, the connected domain information of single connected region；
In every a line of information array, first row being used for storing the starting pixels point place of the single connected domain of current line, second is used for storing the row terminating pixel place of the single connected domain of current line, 3rd number being used for storing the pixel of single connected domain, 4th mark value being used for storing this connected domain, the 5th the line order row being used for storing current line；The mark value of connected domain is an increasing sequence by 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610073505.XA CN105740828B (en) | 2016-02-02 | 2016-02-02 | A kind of stopping line detecting method based on Fast Labeling connection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610073505.XA CN105740828B (en) | 2016-02-02 | 2016-02-02 | A kind of stopping line detecting method based on Fast Labeling connection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105740828A true CN105740828A (en) | 2016-07-06 |
| CN105740828B CN105740828B (en) | 2019-07-19 |
Family
ID=56242254
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610073505.XA Active CN105740828B (en) | 2016-02-02 | 2016-02-02 | A kind of stopping line detecting method based on Fast Labeling connection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105740828B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107301773A (en) * | 2017-06-16 | 2017-10-27 | 上海肇观电子科技有限公司 | A kind of method and device to destination object prompt message |
| CN108647570A (en) * | 2018-04-02 | 2018-10-12 | 深圳市易成自动驾驶技术有限公司 | Zebra line detecting method, device and computer readable storage medium |
| CN110705441A (en) * | 2019-09-27 | 2020-01-17 | 四川长虹电器股份有限公司 | Pedestrian crossing line image post-processing method and system |
| CN111079541A (en) * | 2019-11-19 | 2020-04-28 | 重庆大学 | Road stop line detection method based on monocular vision |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007066003A (en) * | 2005-08-31 | 2007-03-15 | Toyota Motor Corp | Stop-line detection device |
| CN101819681A (en) * | 2009-12-16 | 2010-09-01 | 东南大学 | Weight number adaptively adjusted weighted average background updating method |
| CN103488976A (en) * | 2013-09-17 | 2014-01-01 | 北京联合大学 | Stop mark real-time detection and distance measurement method based on intelligent driving |
| CN104504364A (en) * | 2014-11-23 | 2015-04-08 | 北京联合大学 | Real-time stop line recognition and distance measurement method based on temporal-spatial correlation |
| JP5794023B2 (en) * | 2011-07-28 | 2015-10-14 | アイシン・エィ・ダブリュ株式会社 | Stop line detection system, stop line detection device, stop line detection method, and computer program |
| JP2015222536A (en) * | 2014-05-23 | 2015-12-10 | 日産自動車株式会社 | Marking line detector and marking line detection method |
-
2016
- 2016-02-02 CN CN201610073505.XA patent/CN105740828B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007066003A (en) * | 2005-08-31 | 2007-03-15 | Toyota Motor Corp | Stop-line detection device |
| CN101819681A (en) * | 2009-12-16 | 2010-09-01 | 东南大学 | Weight number adaptively adjusted weighted average background updating method |
| JP5794023B2 (en) * | 2011-07-28 | 2015-10-14 | アイシン・エィ・ダブリュ株式会社 | Stop line detection system, stop line detection device, stop line detection method, and computer program |
| CN103488976A (en) * | 2013-09-17 | 2014-01-01 | 北京联合大学 | Stop mark real-time detection and distance measurement method based on intelligent driving |
| JP2015222536A (en) * | 2014-05-23 | 2015-12-10 | 日産自動車株式会社 | Marking line detector and marking line detection method |
| CN104504364A (en) * | 2014-11-23 | 2015-04-08 | 北京联合大学 | Real-time stop line recognition and distance measurement method based on temporal-spatial correlation |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107301773A (en) * | 2017-06-16 | 2017-10-27 | 上海肇观电子科技有限公司 | A kind of method and device to destination object prompt message |
| CN108647570A (en) * | 2018-04-02 | 2018-10-12 | 深圳市易成自动驾驶技术有限公司 | Zebra line detecting method, device and computer readable storage medium |
| CN108647570B (en) * | 2018-04-02 | 2021-08-06 | 深圳市易成自动驾驶技术有限公司 | Zebra crossing detection method and device and computer readable storage medium |
| CN110705441A (en) * | 2019-09-27 | 2020-01-17 | 四川长虹电器股份有限公司 | Pedestrian crossing line image post-processing method and system |
| CN110705441B (en) * | 2019-09-27 | 2022-11-25 | 四川长虹电器股份有限公司 | Pedestrian crossing line image post-processing method and system |
| CN111079541A (en) * | 2019-11-19 | 2020-04-28 | 重庆大学 | Road stop line detection method based on monocular vision |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105740828B (en) | 2019-07-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Park et al. | Patch-based crack detection in black box images using convolutional neural networks | |
| CN102364496B (en) | Method and system for identifying automobile license plates automatically based on image analysis | |
| CN109284758B (en) | Invoice seal eliminating method and device and computer storage medium | |
| CN108090423B (en) | Depth license plate detection method based on thermodynamic diagram and key point regression | |
| CN101430195B (en) | Method for computing electric power line ice-covering thickness by using video image processing technology | |
| CN104298976B (en) | Detection method of license plate based on convolutional neural networks | |
| CN104008387B (en) | Lane line detection method based on feature point piecewise linear fitting | |
| CN100414561C (en) | Vehicle plate extracting method based on skiagraphy and mathematical morphology | |
| CN109190481B (en) | Method and system for extracting road material of remote sensing image | |
| CN103383733B (en) | A kind of track based on half machine learning video detecting method | |
| CN108053419A (en) | Inhibited and the jamproof multiscale target tracking of prospect based on background | |
| CN107609491A (en) | A kind of vehicle peccancy parking detection method based on convolutional neural networks | |
| CN105740828A (en) | Stop line detection method based on quick sign communication | |
| CN106373426A (en) | Computer vision-based parking space and illegal lane occupying parking monitoring method | |
| CN104021574A (en) | Method for automatically identifying pavement diseases | |
| CN104050450A (en) | Vehicle license plate recognition method based on video | |
| CN103020623A (en) | Traffic sign detection method and equipment | |
| CN103902985B (en) | High-robustness real-time lane detection algorithm based on ROI | |
| CN102708356A (en) | Automatic license plate positioning and recognition method based on complex background | |
| CN104700072A (en) | Lane line historical frame recognition method | |
| CN102254152A (en) | License plate location method based on color change points and color density | |
| CN108241829A (en) | Vehicle travels image-recognizing method | |
| CN106355180B (en) | A kind of license plate locating method combined based on color with edge feature | |
| CN102999749A (en) | Intelligent safety belt regulation violation event detecting method based on face detection | |
| CN108509950B (en) | Railway contact net support number plate detection and identification method based on probability feature weighted fusion |
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 |