CN105046198A - Lane detection method - Google Patents

Lane detection method Download PDF

Info

Publication number
CN105046198A
CN105046198A CN201510325357.1A CN201510325357A CN105046198A CN 105046198 A CN105046198 A CN 105046198A CN 201510325357 A CN201510325357 A CN 201510325357A CN 105046198 A CN105046198 A CN 105046198A
Authority
CN
China
Prior art keywords
lines
line
point
lane
lane line
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
Application number
CN201510325357.1A
Other languages
Chinese (zh)
Other versions
CN105046198B (en
Inventor
江汛洋
黄修源
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
SHANGHAI XIUYUAN NETWORK TECHNOLOGY Co Ltd
Original Assignee
SHANGHAI XIUYUAN NETWORK TECHNOLOGY Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by SHANGHAI XIUYUAN NETWORK TECHNOLOGY Co Ltd filed Critical SHANGHAI XIUYUAN NETWORK TECHNOLOGY Co Ltd
Priority to CN201510325357.1A priority Critical patent/CN105046198B/en
Publication of CN105046198A publication Critical patent/CN105046198A/en
Application granted granted Critical
Publication of CN105046198B publication Critical patent/CN105046198B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/00624Recognising scenes, i.e. recognition of a whole field of perception; recognising scene-specific objects
    • G06K9/00791Recognising scenes perceived from the perspective of a land vehicle, e.g. recognising lanes, obstacles or traffic signs on road scenes
    • G06K9/00798Recognition of lanes or road borders, e.g. of lane markings, or recognition of driver's driving pattern in relation to lanes perceived from the vehicle; Analysis of car trajectory relative to detected road

Abstract

The present invention provides a lane detection method, comprising the implementation steps of: acquiring vehicle front viewing images; intercepting a vehicle front part region of a certain frame image, and naming the vehicle front part region as Roi1; preprocessing the Roi1; performing Hough conversion for the Roi1; obtaining straight lines and center lines; connecting the straight lines with close heads and tails to form a single line; scoring each single line respectively; traversing a CL set, adding scores to lines capable of forming lane lines with lines in an L set, so as to screen out more lines for becoming the lane line; selecting one lane line L in the L set, detecting inner points of the lane line L; fitting a polynomial P according to the inner points; and tracking the polynomial P through a tracking algorithm in the next frame image to predict a lane line. The lane detection method of the present invention is high in identifying accuracy and quick in identifying speed, is capable of identifying curves, and is capable of filtering surface mark noise.

Description

A kind of lane detection method
Technical field
The present invention relates to active safety systems of vehicles field, be specifically related to a kind of lane detection method.
Background technology
Current lane detection is mainly used in the exploitation of intelligent vehicle, and on the vision navigation system being mostly used in intelligent vehicle or driveway deviation alarming system, due to the complicacy of road environment, the research of most of lane detection technology is exactly structure based road.Structured road refers generally to highway and partial structured good highway.Lane line on these highways and road boundary clear, lane line is generally continuous linear, curve or dash line.The lane position of structured road and shape mainly rely on detection traffic lane line and lane boundary to obtain, this is because in a practical situation, lane line on road is easily worn, cause detection difficult, therefore need to detect the position that traffic lane line and lane boundary accurately could determine track simultaneously.
Present stage, the method for lane identification mainly contained two kinds, characteristics of image method and model matching method.The basic thought of characteristics of image method utilizes road boundary or graticule and the difference of other backgrounds on characteristics of image such as house, meadow to carry out Road Detection, these features comprise shape, gray scale, texture, color, contrast and uncontinuity etc., main calculating concentrates on image procossing, and how to carry out the categorization of perception of eigenwert, but the result that factors affect detects such as illumination variation, shade block, noise, road boundary or markings are discontinuous.In addition, many local extremums that characteristics of image provides are only optimal solution, and the correct understanding of track feature also will depend on track model contour level knowledge.The basic thought of model matching method is the priori according to road, utilizes two dimension or three-dimensional curve to carry out road modeling, estimates track model parameter in conjunction with vision mode and characteristics of image.Common two-dimentional track model has the forms such as straight line, SPL and para-curve.Hough method of changing through being commonly used to detect the method in forthright track in model matching method.Hough change can detection of straight lines and junction point on the same line, can be used for detection meets the various curves of analytic expression (x, y)=0 form, and its major advantage make use of image overall characteristic, thus the impact by noise and border interruption is less, strong robustness.
Application number is that the Chinese patent of CN201310534595.4 discloses a kind of real-time lane line detection method, area-of-interest is determined specifically by present frame end point position, remove the first half image not having lane line, shorten the processing time of every two field picture.At area-of-interest from inside to outside, two directions are divided into carry out marginal point scanning, what detect is all inner side apart from the nearest lane line of vehicle on track at every turn, except the interference of other marginal points, shake can not be produced because of lane line width, and due to from end point position by picture dimidiation inspection vehicle diatom, avoid when full figure is detected, two straight lines detected, all in the situation of end point side, improve the accuracy of detection.In addition, when this patented method is with Hough transform lane detection, only gets the straight line in certain angular range with horizontal line, improve the accuracy of detection, save calculating not time of straight line Hough weights in scope.Although this patent can quick and precisely stably realize, on highway and the good rural highway lane detection of road conditions, when running into bend, cannot determining end point, namely algorithm lost efficacy.
Application number is that the Chinese patent of CN201410475019.1 discloses a kind of method for detecting lane lines, comprise the following steps: first pretreated image is laterally divided into K edge image block, and the height ratio of the edge image block of bottom and entire image is [1/4,1/3], below determined vanishing line, two are determined to the above straight line nearest from vanishing line to the candidate's straight line pair as current edge image block; Again based on calculating the right weights of each candidate's straight line respectively, to there is the right edge image block of candidate's straight line, get the straight line of maximum weights to the unique track line segment pair for current edge image block; Finally based on the end points that the track line segment of each edge image block is right, export the lane line of current frame image.This patent is used for lane detection, and it is insensitive to initiation parameter, and the robustness of detection is high, and, lane line disappearance comparatively dark in illumination and all can obtain good Detection results under having the mal-conditions such as shade, but this patent can only be used for detection of straight lines.
Therefore, in the art, be badly in need of a kind of can process bend situation and Detection results is good, recognition speed is fast and the lane detection method that accuracy is high.
Summary of the invention
It is high that technical matters to be solved by this invention is to provide a kind of recognition accuracy rate, and recognition speed is fast, can identify bend, can filter the lane detection method of surface mark noise.
For solving the problems of the technologies described above, the technical solution used in the present invention is:
A kind of lane detection method, is characterized in that, comprise the following steps.
First step: vehicle camera obtains vehicle front view image continuously.
Second step: the region, front part intercepting a certain two field picture, called after Roi1.
Third step: carry out pre-service to Roi1, it comprises the cromogram of Roi1 is converted to gray-scale map, converts the image of Roi1 to vertical view by IPM conversion, uses canny algorithm to the level and smooth noise of Roi1, smooth edges, and edge calculation figure.
Preferably, pretreatment module can carry out pre-service to Roi1.
4th step: do Hough transform to Roi1, obtains straight line.
Preferably, Hough transformation module can realize the process of the 4th step.
5th step: obtain all close disjoint straight lines between two, abandons non-close disjoint straight line between two, and obtains the center line of close non-intersect straight line between two, obtains all straight lines with brightness, abandon the straight line in center line without brightness in center line.
Preferably, computing module can realize the process of the 5th step.
6th step: the close straight line of head and the tail is connected into wall scroll lines, and by described lines collection called after CL collection.
7th step: have certain distance between lane line to lane line and keep similar slope, lines described in every bar are scored separately, to adjust the distance lines bonus point that the is close and line features that slope is similar, to the lines deduction not having described line features, then the lines of the close positions in next frame image are retrieved, to the lines bonus point with described line features, to the lines deduction not having described line features.
Preferably, the process that sub-module can realize the 7th step is added.
8th step: repeated execution of steps S7 until occur that score exceedes the lines of lane line threshold value, then thinks that these lines may be lane line, by this lines collection called after L collection.
9th step: L concentrates all lines slopes not conflict mutually, then think that current vehicle diatom estimates, otherwise then repeats step S7, until all lines slopes that L concentrates do not conflict mutually.
Tenth step: traversal CL collection, the lines can concentrated with L are formed to the lines bonus point of lane line, namely be the lines distance concentrated to L at certain threshold range and the similar lines at certain threshold range of slope, lines score being exceeded lane line threshold value are put into L and are concentrated.
11 step: choose one article of lane line L that L concentrates, the interior point of inspection vehicle diatom L, point should have and in gray-scale map, have certain brightness and in Roi1 outline map and be the point patterns of the point on 2 straight line center lines in lane line.
12 step: put set on lane line L, carry out fitting of a polynomial X=a 0+ a 1y+a 2y 2+ a 3y 3, obtain polynomial expression P.
Preferably, computing module can realize the process of the 12 step.
13 step: in next frame image, uses track algorithm to follow the tracks of polynomial expression P, predict lane line, concentrates the lane line found near predict lane line at CL, and according to the lane line location updating polynomial expression P found.
Preferably, computing module can realize the process of the 8th step to the 13 step.
Further, following the tracks of polynomial expression P and find lane line in next frame image after, then to this lane line bonus point corresponding in current frame image, and controlling the constant that score is no more than a setting; If do not find lane line, then to this lane line deduction corresponding in current frame image.
1. further, in the 4th step, when doing Hough transform to Roi1, if straight line quantity is less than predetermined threshold value, then first use high cap computing, re-use Hough transform and obtain straight line.
1. further, in described 11 step, the process of the interior point of inspection vehicle diatom L is: the path along lane line L solves interior point from bottom to top, linear equation y=kx+b is created according to straight line on L, from straight line starting point, from a distance in y-axis, x coordinate is calculated according to linear equation, and find the point with point patterns in lane line at this x nearby coordinates, the point with point patterns in lane line is found according to starting point and described x nearby coordinates, upgrade the k in linear equation, the value of b, and again there is the point of point patterns in lane line for starting point with described x nearby coordinates searching, continue to try to achieve next point in conjunction with described linear equation, double counting, until can not find the point with point patterns in lane line, now get lower straight line in L again, continue to calculate, until can not find the point with point patterns in lane line and the straight line exhausted in L.
Particularly, in described 11 step, the process of the interior point of inspection vehicle diatom L is:
1.L is that many line segments are formed by connecting, and is set to (l 0, l 1, l 2, ..., l n)
2. make i=0
3.l ifor linear equation (can y=kx+b be expressed as), calculate k, b, get l istarting point, (x 1, y 1)
4.y 2=y 1-d, (d is constant)
5. if y 2lower than l i+1starting point, then make i=i+1, and jump to 3
6. if y 2arrive bottom Roi1, then terminate to calculate
7.x 2=(y 2+ b)/k, at x 2neighbouring searching meets the point of point patterns in lane line.
8. according to (x 1, y 1), (x 2, y 2) new linear equation can be obtained, upgrade k, b according to this equation.
Make x 2=x 1, y2=y 1and jump to 4.
In described 11 step, luminance threshold computing method are as follows, to pick up the car lines upper extreme point brightness on diatom L to deduct value be that constant between 20 to 30 obtains c1, if environs detects lane line in previous frame, its threshold value c0, then to c1, c0 weighted mean, try to achieve final threshold value, otherwise then directly to get c1 be threshold value.
Further, the track algorithm in described 13 step is Kalman filtering.
Further, in described second step, also comprise after processing a frame road image, utilize previous frame Lane detection result, dynamically adjust interested regional extent, the region at lane line place in prediction present frame, and then limit the region of search of image in current identifying, carry out the search of lane line next time.
Further, in described third step, after acquisition road gray level image, in order to make lane line and non-lane line separately, needing to carry out Iamge Segmentation process, choosing suitable image segmentation threshold.
Further, at described second step, intercept the region, front part of a certain two field picture, after called after Roi1, the level and smooth noise of Neighborhood Filtering, smooth edges can also be used, and use edge detection algorithm edge calculation figure.Field filtering square frame, Gauss, intermediate value, bilateral filtering; Edge detection algorithm comprises RobertsCross operator, Prewitt operator, Sobel operator, compass operator.
Compared with prior art, the invention has the beneficial effects as follows that lane identification accuracy rate is high, recognition speed is fast, can identify bend, can filter surface mark noise.
Accompanying drawing explanation
Fig. 1 is the schematic flow sheet of a kind of lane detection method according to the first embodiment of the present invention.
Fig. 2 overlooks gray-scale map according to the track of a kind of lane detection method of the first embodiment of the present invention.
Fig. 3 is the track obfuscation outline map of a kind of lane detection method according to the first embodiment of the present invention.
Fig. 4 be according to a kind of lane detection method of the first embodiment of the present invention with the outline map put in lane line.
Fig. 5 is the polynomial fitting equation schematic diagram of a kind of lane detection method according to the first embodiment of the present invention.
Fig. 6 is the schematic flow sheet of a kind of lane detection method according to a second embodiment of the present invention.
Embodiment
Below in conjunction with the drawings and specific embodiments, the present invention is further illustrated.The object of this invention is to provide a kind of recognition accuracy rate high, recognition speed is fast, can identify bend, can filter the lane detection method of surface mark noise.
Embodiment 1:
The schematic flow sheet of a kind of lane detection method according to the first embodiment of the present invention as shown in Figure 1.See Fig. 1, a kind of lane detection method of the present invention comprises the following steps.
First step S1: vehicle camera obtains vehicle front view image continuously.
Second step S2: the region, front part intercepting a certain two field picture, called after Roi1.
Third step S3: pre-service is carried out to Roi1, it comprises and uses method of average coloured image being converted into gray level image that the cromogram of Roi1 is converted to gray-scale map, convert the image of Roi1 to vertical view by IPM conversion, use canny algorithm to Roi1 to the level and smooth noise of Roi1, smooth edges and edge calculation figure.Wherein, gray-scale map is overlooked in track as shown in Figure 2, and coloured image is converted into gray level image by (R+G+B)/3, is converted to gray-scale map by Roi1, Roi1 is converted to vertical view by IPM conversion (skeleton view is turned into vertical view by inverse perspective mapping).Track obfuscation outline map as shown in Figure 3, canny algorithm is to Roi1 edge calculation figure.
4th step S4: do Hough transform to Roi1, obtains straight line.
5th step S5: obtain all close disjoint straight lines between two, abandons non-close disjoint straight line between two, and obtains the center line of close non-intersect straight line between two, obtains all straight lines with brightness, abandon the straight line in center line without brightness in center line.
6th step S6: the close straight line of head and the tail is connected into wall scroll lines, and by described lines collection called after CL collection.
7th step S7: have certain distance between lane line to lane line and keep similar slope, described in the every bar concentrate CL, lines are scored separately, to adjust the distance lines bonus point that the is close and line features that slope is similar, to the lines deduction not having described line features, then the lines of the close positions in next frame image are retrieved, to the lines bonus point with described line features, to the lines deduction not having described line features.
8th step S8: repeated execution of steps S7 until occur that score exceedes the lines of lane line threshold value, then think that these lines may be lane line, by this lines collection called after L collection.
9th step S9:L concentrates all lines slopes not conflict mutually, then think that current vehicle diatom estimates, otherwise then repeats step S7, until all lines slopes that L concentrates do not conflict mutually.
Tenth step S10: choose one article of lane line L that L concentrates, the interior point of inspection vehicle diatom L, point should have and in gray-scale map, have certain brightness and in Roi1 outline map and be the point patterns of the point on 2 straight line center lines in lane line, the outline map of point in lane line of the present invention shown in Figure 4.
11 step S11: put set on lane line L, carry out fitting of a polynomial X=a 0+ a 1y+a 2y 2+ a 3y 3, obtain polynomial expression P, fitting of a polynomial effect is see polynomial fitting equation schematic diagram of the present invention as shown in Figure 5.
12 step S12: in next frame image, uses track algorithm to follow the tracks of polynomial expression P, predict lane line, concentrates the lane line found near predict lane line at CL, and according to the lane line location updating polynomial expression P found; If found, then increase this lane line score, and control score and be no more than setting constant; If can not find, reduce score, score is reduced to lower than threshold value, then think that this lane line disappears.When road surface, lane line feature is not obvious, or because barrier blocks, or because of process of passing through tunnel, now reduce score, make predict lane line be unlikely to disappear immediately, when again finding lane line in original image, progressively can increase score, and reaching certain numerical value.
By the scheme of embodiment 1, lane detection method recognition accuracy rate of the present invention is high, and recognition speed is fast, can identify bend, can filter surface mark noise.In the lane detection process of reality, when Hough transform acquisition straight line is done to Roi1, because straight line quantity is likely less than predetermined threshold value, in order to the present embodiment has better effect, then first use high cap computing, re-use Hough transform and obtain straight line.Such as, when crossing tunnel, because visibility is low, using morphologic high cap to calculate, making bright spot brighter, thus improve visibility, re-use Hough transform, both taken full advantage of the information of image, ensure that again the robustness of algorithm to a certain extent.
Embodiment 2:
The schematic flow sheet of a kind of lane detection method according to a second embodiment of the present invention as shown in Figure 6.See Fig. 6, a kind of lane detection method of the present invention comprises the following steps.
First step S1: vehicle camera obtains vehicle front view image continuously.
Second step S2: the region, front part intercepting a certain two field picture, called after Roi1; Intercept the region, front part of a certain two field picture, after called after Roi1, use canny algorithm to image smoothing noise, smooth edges and edge calculation figure.
Third step S3: pre-service is carried out to Roi1, it comprises and uses method of average coloured image being converted into gray level image that the cromogram of Roi1 is converted to gray-scale map, convert the image of Roi1 to vertical view by IPM conversion, use canny algorithm to Roi1 smooth noise, smooth edges edge calculation figure.
4th step S4: do Hough transform to Roi1, obtains straight line.
5th step S5: obtain all close disjoint straight lines between two, abandons non-close disjoint straight line between two, and obtains the center line of close non-intersect straight line between two, obtains all straight lines with brightness, abandon the straight line in center line without brightness in center line.
6th step S6: the close straight line of head and the tail is connected into wall scroll lines, and by described lines collection called after CL collection.
7th step S7: have certain distance between lane line to lane line and keep similar slope, described in the every bar concentrate CL, lines are scored separately, to adjust the distance lines bonus point that the is close and line features that slope is similar, to the lines deduction not having described line features, then the lines of the close positions in next frame image are retrieved, to the lines bonus point with described line features, to the lines deduction not having described line features.
8th step S8: repeated execution of steps S7 until occur that score exceedes the lines of lane line threshold value, then think that these lines may be lane line, by this lines collection called after L collection.
9th step S9:L concentrates all lines slopes not conflict mutually, then think that current vehicle diatom estimates, otherwise then repeats step S7, until all lines slopes that L concentrates do not conflict mutually.
Tenth step S10: choose one article of lane line L that L concentrates, the interior point of inspection vehicle diatom L, point should have and in gray-scale map, have certain brightness and in Roi1 outline map and be the point patterns of the point on 2 straight line center lines in lane line.
11 step S11: put set on lane line L, carry out fitting of a polynomial X=a 0+ a 1y+a 2y 2+ a 3y 3, obtain polynomial expression P.
12 step S12: in next frame image, uses track algorithm to follow the tracks of polynomial expression P, predict lane line, concentrates the lane line found near predict lane line at CL, and according to the lane line location updating polynomial expression P found.
Embodiment is only the conventional embodiment of the present invention above.Obviously, various supplement, amendment and replacement can be had under the prerequisite not departing from the present invention's spirit that claims define and invention scope.It should be appreciated by those skilled in the art that the present invention according to concrete environment and job requirement under the prerequisite not deviating from invention criterion, can change in form, layout, element and other side in actual applications to some extent.Therefore, be only illustrative rather than definitive thereof in the embodiment of this disclosure, the scope of the present invention is defined by claim and legal equivalents thereof, and is not limited thereto front description.

Claims (7)

1. a lane detection method, is characterized in that, comprises the following steps:
First step (S1): vehicle camera obtains vehicle front view image continuously;
Second step (S2): the region, front part intercepting a certain two field picture, called after Roi1;
Third step (S3): pre-service is carried out to Roi1, it comprises the cromogram of Roi1 is converted to gray-scale map, convert the image of Roi1 to vertical view by IPM conversion, and use canny algorithm to the level and smooth noise of Roi1, smooth edges and edge calculation figure;
4th step (S4): do Hough transform to Roi1, obtains straight line;
5th step (S5): obtain all close disjoint straight lines between two, abandon non-close disjoint straight line between two, and obtain the center line of close non-intersect straight line between two, in center line, obtain all straight lines with brightness, abandon the straight line in center line without brightness;
6th step (S6): the close straight line of head and the tail is connected into wall scroll lines, and by described lines collection called after CL collection;
7th step (S7): have certain distance between lane line to lane line and keep similar slope, described in the every bar concentrate CL, lines are scored separately, to adjust the distance lines bonus point that the is close and line features that slope is similar, to the lines deduction not having described line features, then the lines of the close positions in next frame image are retrieved, to the lines bonus point with described line features, to the lines deduction not having described line features;
8th step (S8): repeat the 7th step (S7) until occur that score exceedes the lines of lane line threshold value, then think that these lines may be lane line, by this lines collection called after L collection;
9th step (S9): L concentrates all lines slopes not conflict mutually, then think that current vehicle diatom estimates, otherwise then repeats the 7th step (S7), until all lines slopes that L concentrates do not conflict mutually;
Tenth step (S10): calculate point in lane line: choose one article of lane line L that L concentrates, the interior point of inspection vehicle diatom L, point should have and in gray-scale map, have certain brightness and in Roi1 outline map and be the point patterns of the point on 2 straight line center lines in lane line;
11 step (S11): put set on lane line L, carry out fitting of a polynomial X=a 0+ a 1y+a 2y 2+ a 3y 3, obtain polynomial expression P;
12 step (S12): in next frame image, uses track algorithm to follow the tracks of polynomial expression P, predict lane line, concentrates the lane line found near predict lane line at CL, and according to the lane line location updating polynomial expression P found.
2. a kind of lane detection method as claimed in claim 1, is characterized in that, in the 4th step (S4), when Hough transform is done to Roi1, if straight line quantity is less than predetermined threshold value, then first use high cap computing, re-use Hough transform and obtain straight line.
3. a kind of lane detection method as claimed in claim 1, it is characterized in that, in described tenth step (S10), the process of the interior point of inspection vehicle diatom L is: the path along lane line L solves interior point from bottom to top, linear equation y=kx+b is created according to straight line on L, from straight line starting point, from a distance in y-axis, x coordinate is calculated according to linear equation, and find the point with point patterns in lane line at this x nearby coordinates, the point with point patterns in lane line is found according to starting point and described x nearby coordinates, upgrade the k in linear equation, the value of b, and again there is the point of point patterns in lane line for starting point with described x nearby coordinates searching, continue to try to achieve next point in conjunction with described linear equation, double counting, until can not find the point with point patterns in lane line, now get lower straight line in L again, continue to calculate, until can not find the point with point patterns in lane line and the straight line exhausted in L.
4. a kind of lane detection method as claimed in claim 1, is characterized in that, the track algorithm in described 12 step (S12) is Kalman filtering.
5. a kind of lane detection method as claimed in claim 1, it is characterized in that, in described tenth step, luminance threshold computing method are, lines upper extreme point brightness on the diatom L that picks up the car to deduct value be that constant between 20 to 30 obtains c1, if environs detects lane line in previous frame, its threshold value c0, then to c1, c0 weighted mean, try to achieve final threshold value, otherwise then directly to get c1 be threshold value.
6. a kind of lane detection method according to any one of claim 1-5, it is characterized in that, at described second step (S2), intercept the region, front part of a certain two field picture, after called after Roi1, field filtering can also be used image smoothing noise, smooth edges, and use edge detection algorithm edge calculation figure.
7. a kind of lane detection method as claimed in claim 1, it is characterized in that, in the 9th step, traversal CL collection, the lines can concentrated with L are formed to the lines bonus point of lane line, namely be the lines distance concentrated to L at certain threshold range and the similar lines at certain threshold range of slope, lines score being exceeded lane line threshold value are put into L and are concentrated.
CN201510325357.1A 2015-06-12 2015-06-12 A kind of lane detection method Expired - Fee Related CN105046198B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510325357.1A CN105046198B (en) 2015-06-12 2015-06-12 A kind of lane detection method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510325357.1A CN105046198B (en) 2015-06-12 2015-06-12 A kind of lane detection method

Publications (2)

Publication Number Publication Date
CN105046198A true CN105046198A (en) 2015-11-11
CN105046198B CN105046198B (en) 2018-07-06

Family

ID=54452727

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510325357.1A Expired - Fee Related CN105046198B (en) 2015-06-12 2015-06-12 A kind of lane detection method

Country Status (1)

Country Link
CN (1) CN105046198B (en)

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105426868A (en) * 2015-12-10 2016-03-23 山东大学 Lane detection method based on adaptive region of interest
CN105608429A (en) * 2015-12-21 2016-05-25 重庆大学 Differential excitation-based robust lane line detection method
CN105956511A (en) * 2016-04-18 2016-09-21 江苏大学 Lane line detecting and combining method based on Hough transform
CN107392080A (en) * 2016-05-17 2017-11-24 纬创资通股份有限公司 Image quided method and its electronic installation
CN107832734A (en) * 2017-11-24 2018-03-23 天津大学 A kind of method for detecting lane lines based on the extraction of three phase characteristics
CN107909047A (en) * 2017-11-28 2018-04-13 上海信耀电子有限公司 A kind of automobile and its lane detection method and system of application
CN108171225A (en) * 2018-03-14 2018-06-15 海信集团有限公司 Lane detection method, device, terminal and storage medium
CN109591694A (en) * 2017-09-30 2019-04-09 上海欧菲智能车联科技有限公司 Lane Departure Warning System, lane departure warning method and vehicle
CN111731324A (en) * 2020-05-29 2020-10-02 徐帅 Control method and system for guiding AGV intelligent vehicle based on vision
CN109591694B (en) * 2017-09-30 2021-09-28 上海欧菲智能车联科技有限公司 Lane departure early warning system, lane departure early warning method and vehicle

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130120125A1 (en) * 2011-11-16 2013-05-16 Industrial Technology Research Institute Method and system for lane departure warning
CN103383733A (en) * 2013-05-16 2013-11-06 浙江智尔信息技术有限公司 Lane video detection method based on half-machine study
CN103593649A (en) * 2013-10-24 2014-02-19 惠州华阳通用电子有限公司 Lane line detection method for lane departure early warning
CN103617412A (en) * 2013-10-31 2014-03-05 电子科技大学 Real-time lane line detection method
CN104268860A (en) * 2014-09-17 2015-01-07 电子科技大学 Lane line detection method

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130120125A1 (en) * 2011-11-16 2013-05-16 Industrial Technology Research Institute Method and system for lane departure warning
CN103383733A (en) * 2013-05-16 2013-11-06 浙江智尔信息技术有限公司 Lane video detection method based on half-machine study
CN103593649A (en) * 2013-10-24 2014-02-19 惠州华阳通用电子有限公司 Lane line detection method for lane departure early warning
CN103617412A (en) * 2013-10-31 2014-03-05 电子科技大学 Real-time lane line detection method
CN104268860A (en) * 2014-09-17 2015-01-07 电子科技大学 Lane line detection method

Cited By (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105426868B (en) * 2015-12-10 2018-09-28 山东大学 A kind of lane detection method based on adaptive area-of-interest
CN105426868A (en) * 2015-12-10 2016-03-23 山东大学 Lane detection method based on adaptive region of interest
CN105608429A (en) * 2015-12-21 2016-05-25 重庆大学 Differential excitation-based robust lane line detection method
CN105608429B (en) * 2015-12-21 2019-05-14 重庆大学 Robust method for detecting lane lines based on difference excitation
CN105956511A (en) * 2016-04-18 2016-09-21 江苏大学 Lane line detecting and combining method based on Hough transform
CN105956511B (en) * 2016-04-18 2019-04-02 江苏大学 A method of lane straight-line detection based on Hough transform with merge
CN107392080A (en) * 2016-05-17 2017-11-24 纬创资通股份有限公司 Image quided method and its electronic installation
CN109591694B (en) * 2017-09-30 2021-09-28 上海欧菲智能车联科技有限公司 Lane departure early warning system, lane departure early warning method and vehicle
CN109591694A (en) * 2017-09-30 2019-04-09 上海欧菲智能车联科技有限公司 Lane Departure Warning System, lane departure warning method and vehicle
CN107832734A (en) * 2017-11-24 2018-03-23 天津大学 A kind of method for detecting lane lines based on the extraction of three phase characteristics
CN107909047A (en) * 2017-11-28 2018-04-13 上海信耀电子有限公司 A kind of automobile and its lane detection method and system of application
CN108171225A (en) * 2018-03-14 2018-06-15 海信集团有限公司 Lane detection method, device, terminal and storage medium
CN111731324A (en) * 2020-05-29 2020-10-02 徐帅 Control method and system for guiding AGV intelligent vehicle based on vision

Also Published As

Publication number Publication date
CN105046198B (en) 2018-07-06

Similar Documents

Publication Publication Date Title
CN105046198A (en) Lane detection method
USRE48106E1 (en) Detection of obstacles at night by analysis of shadows
JP5874756B2 (en) Marking line detection system and marking line detection method
US9569675B2 (en) Three-dimensional object detection device, and three-dimensional object detection method
US10262216B2 (en) Hazard detection from a camera in a scene with moving shadows
US9591274B2 (en) Three-dimensional object detection device, and three-dimensional object detection method
US9141870B2 (en) Three-dimensional object detection device and three-dimensional object detection method
CN104102905B (en) A kind of adaptive detection method of lane line
KR20150102546A (en) Apparatus and method for recognizing lane
US9558546B2 (en) Three-dimensional object detection device, and three-dimensional object detection method
CN102789234A (en) Robot navigation method and robot navigation system based on color coding identifiers
JP5776795B2 (en) Three-dimensional object detection device
CN106529443A (en) Method for improving detection of lane based on Hough transform
US20150002672A1 (en) Three-dimenisional object detection device
CN104916163A (en) Parking space detection method
US9589193B2 (en) Three-dimensional object detection device and three-dimensional object detection method
US20150178575A1 (en) Three-dimensional object detection device and three-dimensional object detection method
JP4946175B2 (en) Road boundary detection device and road boundary detection method
CN105303160A (en) Method for detecting and tracking vehicles at night
Li et al. Nighttime lane markings recognition based on Canny detection and Hough transform
US9832444B2 (en) Three-dimensional object detection device
JP2012150595A (en) Stop line detection device and program
Zhu et al. A method of lane detection and tracking for expressway based on ransac
JP6337601B2 (en) Three-dimensional object detection device
CN110379168B (en) Traffic vehicle information acquisition method based on Mask R-CNN

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
CF01 Termination of patent right due to non-payment of annual fee
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20180706

Termination date: 20200612