CN102436644A - Unstructured road detection method based on adaptive edge registration - Google Patents
Unstructured road detection method based on adaptive edge registration Download PDFInfo
- Publication number
- CN102436644A CN102436644A CN2011103414791A CN201110341479A CN102436644A CN 102436644 A CN102436644 A CN 102436644A CN 2011103414791 A CN2011103414791 A CN 2011103414791A CN 201110341479 A CN201110341479 A CN 201110341479A CN 102436644 A CN102436644 A CN 102436644A
- Authority
- CN
- China
- Prior art keywords
- edge
- road
- otsu
- canny
- point
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 32
- 230000003044 adaptive effect Effects 0.000 title claims abstract description 12
- 238000000034 method Methods 0.000 claims abstract description 60
- 238000005457 optimization Methods 0.000 claims abstract description 4
- 230000009977 dual effect Effects 0.000 claims description 10
- 239000004568 cement Substances 0.000 claims description 6
- 230000001788 irregular Effects 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 239000010426 asphalt Substances 0.000 claims description 2
- 230000001105 regulatory effect Effects 0.000 claims description 2
- 230000011218 segmentation Effects 0.000 abstract description 26
- 230000008901 benefit Effects 0.000 abstract description 8
- 238000002474 experimental method Methods 0.000 abstract description 5
- 238000005286 illumination Methods 0.000 abstract description 3
- 238000003708 edge detection Methods 0.000 abstract description 2
- 230000007547 defect Effects 0.000 abstract 1
- 238000012360 testing method Methods 0.000 description 9
- 238000003709 image segmentation Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000005520 cutting process Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 241000408659 Darpa Species 0.000 description 1
- 206010057315 Daydreaming Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 230000001149 cognitive effect Effects 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 230000007123 defense Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000007789 sealing Methods 0.000 description 1
- 238000013179 statistical model Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Images
Abstract
The invention discloses an unstructured road detection method based on adaptive edge registration. The method is the registering method of an Otsu edge and a weighted Canny edge. The method comprises the following steps: weighted Canny edge detection, Otsu threshold optimization and weight revaluation of the Canny edge. Compared to the prior art, the unstructured road detection method based on the adaptive edge registration of the invention has the following advantages that: the registration of the Otsu edge and the weighted Canny edge is taken as a core so that precision of road area segmentation and boundary tracking can be increased; unstructured road identification experiments in different scenes show that influences of unfavorable factors, such as a road defect, a shadow, illumination change and the like can be effectively overcome by using the detection method; the method has good practicality and good economic benefits and social benefits can be generated by using the method.
Description
Technical field
The present invention relates to a kind of destructuring Road Detection method that is used for the intelligent vehicle independent navigation, be specifically related to a kind of destructuring Road Detection method based on the adaptive edge registration.
Background technology
Destructuring Road Detection based on vision is one of focus of intelligent vehicle independent navigation research.Comparing with the air navigation aid based on multi-sensor information fusion, is that the vision navigation method of core provides more cheap solution undoubtedly with the Road Detection.Grand Challenge intelligent vehicle rallisport challenge with advanced studies regional planning agency of present U.S. Department of Defense (DARPA) tissue is an example; The vehicle of taking part in game has mostly been installed expensive non-visual sensor system; Its cost is considerably beyond automobile itself, and this has buried hidden danger for later the research in the application of civil area.The vision guided navigation technology still can't make intelligent vehicle really be independent of the intervention of human pilot owing to receive the restriction of machine vision and cognitive techniques development level.Yet, judge the state of intelligent vehicle run-off-road and prompting in time can effectively avoid because all kinds of traffic hazards of the absent minded initiation of driver (account for total traffic hazard 80%) through Road Detection.
Present road is broadly divided into two types: structured road and destructuring road.Wherein structured road is often referred to highway or the higher highway of rank, and this road has lines sign or regular boundary shape clearly.The destructuring road is meant that generally non-trunk track, city and other do not have the road type (like campus, residential quarter, backroad etc.) of obvious lines sign.Problems such as the destructuring road has out-of-shape, do not have lines sign, road surface to have breakage easily and slight crack, local color or textural characteristics are inhomogeneous and the shadow influence is serious.
The method that detects towards the destructuring road vision is broadly divided into three types: roadway characteristic method, road model method and neural net method.The roadway characteristic method mainly detects road through some characteristic (like color, gray scale, texture, edge or the frequency domain character etc. of road) of extraction and location road.This method utilizes tagsort to realize that major advantage is insensitive to road shape to the cutting apart of road area and non-road area, and the priori that needs is few; Shortcoming is comparatively responsive to shadow, slight crack etc., and computational processing is bigger.Road model method (like straight line model, B-Snake curve model and statistical model etc.) is then set up the parameter model of road in advance, confirms model parameter through graphical analysis, thereby obtains the full detail about road.The detected road area of these class methods is comparatively complete, but can't set up model accurately for the road pavement form of complicacy.Neural network method is utilized the learning characteristic of neural network, does not need the priori of road, but the sample that the result of road Identification adopted when depending on training needs a large amount of training sets.
The vision-based detection problem of destructuring road can be summed up as the image segmentation problem of machine vision.Be directed against image segmentation problem at present and proposed a lot of methods, wherein threshold method does not need extensively employings of advantage quilt such as manual intervention because of it calculates simply.Threshold method is to utilize target and the difference of background on characteristic in the image, carries out the method for image segmentation through choosing appropriate threshold.The most representative threshold method has the expansion on two dimension, multidimensional of Otsu threshold method (maximum variance between clusters), maximum entropy method (MEM), grey level histogram method, minimal error method and each class methods.The application that said method is cut apart at single threshold is more, and opening up of having is wide in many Threshold Segmentation field.
Summary of the invention
Goal of the invention: to deficiency of the prior art; The purpose of this invention is to provide a kind of destructuring Road Detection method based on the adaptive edge registration; To realize that improving road area cuts apart the precision of following the tracks of with the border, effectively overcomes the influence of unfavorable factors such as road is damaged, shadow, illumination change.
Technical scheme: in order to realize the foregoing invention purpose, the technical scheme that the present invention adopts is:
To the characteristics of existing destructuring road, should satisfy following reasonable terms: 1) the feature difference condition of road area and non-road area: mainly refer to the difference on the gray scale in the present invention; 2) " homogeneity " condition of road: no matter be the muddy and ruthed lane and the gravel road of cement road, asphalt road or country, most of zone of road area all has the characteristics of gradation uniformity and gradually changeable; 3) specific region in the image can be regarded as road area: this condition provides possibility for the characteristic information of roughly judging road area, at present the method for existing a lot of physics realizations.
A kind of destructuring Road Detection method based on the adaptive edge registration; Method for registering for Otsu edge and weighting Canny edge; The Otsu method is as a kind of adaptive threshold searching method; The roughly segmentation result of road area and non-road area can only be obtained, and accurate road boundary can't be obtained.Because human cognition for objective things depends on the profile and the shape of things to a great extent except characteristics such as color, gray scale, texture.Therefore particular edge can be considered road and off-highroad accurate boundary.The present invention obtains accurate road boundary by means of the registration at Otsu edge and weighting Canny edge, thereby Region Segmentation is effectively combined with rim detection, guarantees the robustness and the accuracy of algorithm.Specific as follows:
(1) weighting Canny rim detection
Edge detection operator such as Roberts operator, Sobel operator, Laplacian operator, LOG operator, the Canny operator of a lot of classics etc. have been proposed at present.Wherein the Canny operator is to optimize operator a multistage with filtering, enhancing and detection, and the edge of extraction is more complete, and the position is more accurate, can detect the thinner marginal portion of image.Therefore adopt the Canny operator to realize the function of edge extracting.
In traditional C anny operator; Original image
at first obtains the Gaussian Blur image with gaussian kernel
convolution; Then with the finite difference of single order local derviation the assign to amplitude and the direction of compute gradient, gradient magnitude is carried out that non-maximum value suppresses and with detection of dual threshold algorithm and adjoining edge.
At the dual threshold detection-phase, record is identified as the gradient magnitude of edge pixel, so that can unite the irregular border that characterizes the destructuring road with length, angle, distance and the average amplitude intensity of broken line;
At the dual threshold detection-phase, the marginal point that communicates with each other is unified numbering, and after numbering is accomplished, utilize least square method to carry out fitting a straight line, obtain the axis of edge point set with identical numbering;
The equation of supposing straight line does
L:
, then more concentrated any point (
x,
y) to the distance of straight line do
,
Be the intersection point coordinate.The intersection point point
of postulated point
on straight line L is two end points of all intersection point points, can be easy to proof
and set up.So; The compressing mapping that is mapped as of
to
; Mean on line segment
more arbitrarily; All can put on the edge of at least to concentrate and find some correspondence with it; The minimum point of selected distance is its corresponding point, and other points that then have identical intersection point are noncorresponding points.Utilize formula (1) to any point in the edge point set
iCompose weights, wherein S is always counting of edge point set;
Be line segment
The number of last pixel,
L Th Be line of shortest length segment length threshold value (relevant with the resolution of image, as rule of thumb to choose).In formula (1); The weights of edge pixel are directly proportional with
; This means one independently, the little broken line of radian more can obtain bigger weights than the curve sealing with same projection length, that radian is big;
means that then the projection of segment of curve on its axis is long more; Its weights also can be big more, and this has guaranteed that long edge line with having better streamline shape has and bigger possibly be identified as road boundary.
is scale factor; Span is 0 ~ 1, is used for the influence degree of adjusting
to weights.
(2) the threshold value optimization of Otsu
The basic thought of Otsu edge registration is: the accurate boundary of road area and non-road area must be positioned on the Canny edge of image.Therefore, utilize weighting Canny edge can the threshold value of Otsu be optimized, obtain accurate segmentation threshold.
The step of Otsu edge registration is following:
1) all the Canny marginal points in the image is calculated the weighting grey level histogram
, wherein
For gray scale does
iThe number of marginal point,
Be pixel
Weights.
2) in the Otsu threshold value
T Best
Near search for, choose the weighting grey level histogram
Middle distance
T Best Nearest maximum point is as the new threshold value of Otsu algorithm.The decision criteria of maximum point does
, wherein
For nearest maximum point arrives
T Best Distance.
(3) the weights revaluation at Canny edge
Dual threshold at traditional C anny edge is chosen the stage; The threshold value that reduces gradient magnitude can increase the quantity that the edge detects; The weak edge that feasible part belongs to road boundary can keep, but brings higher computation burden can for the fitting a straight line or the curve fitting of back; The threshold value that increases gradient magnitude can reduce the quantity that the edge detects, but the weak edge that may cause belonging to road boundary disappears.In order to address this problem; We select for use less threshold value to keep weak edge on the one hand; The road and the non-road area that utilize the Otsu threshold method to cut apart on the one hand carry out weights distribution and filtering to the Canny edge; When eliminating the complex edge interference, keep near also Canny edge, enhancing Otsu edge (boundary between two types).
Formula (2) is an edge weights revaluation formula; Wherein
is weighting Canny edge image; Locating to get at 0 o'clock at
is non-edge pixel, and value was an edge pixel more than or equal to 1 o'clock;
is the Otsu image after optimizing; Getting at 1 o'clock is road area, gets 0 or be non-road area at-1 o'clock.Formula (2) is to being that corresponding point on the diagonal of a matrix of 2r*2r at center are judged with the Canny edge.(Otsu edge) this marginal point Canny edge will be able to keep when occurring differing from the value of neighbor pixel on the diagonal line, and wherein
is the xor operation symbol.
After the revaluation through the Canny edge, being positioned at the inner part edge of target and background will be eliminated, and the edge that is positioned at the target and background intersection will be able to keep and strengthen.Can utilize edge filter to reduce the calculated amount of back straight line or curve fitting on the one hand on the one hand through reducing the quantity that the increase of gradient magnitude threshold value can detect the edge like this.
Beneficial effect: compared with prior art; The outstanding advantage of the destructuring Road Detection method based on the adaptive edge registration of the present invention comprises: the present invention is a core with the registration at Otsu edge and weighting Canny edge; Improve road area and cut apart the precision of following the tracks of with the border; Destructuring road Identification experiment under the different scenes shows; This detection method can effectively overcome the influence of unfavorable factors such as road is damaged, shadow, illumination change, has good practicability, can produce good economic benefits and social benefit.
Description of drawings
Fig. 1 is the testing result of the sand-gravel surface of embodiment 1, and left side figure is an original image, and right figure is the Canny edge image;
Fig. 2 is the testing result of the sand-gravel surface of embodiment 1, and left side figure is the segmentation result before the registration, and right figure is the segmentation result behind the registration;
Fig. 3 is the testing result of the sand-gravel surface of embodiment 1, and left figure and right figure are respectively the segmentation threshold before and after the registration;
Fig. 4 is the testing result of the sand-gravel surface of embodiment 1, and left figure and right figure are respectively the road segmentation result based on maximum entropy threshold method and standard Otsu threshold method;
Fig. 5 is the testing result on the earth road surface of embodiment 2, and left side figure is the original image of muddy and ruthed lane, and right figure is the road segmentation result of edge method for registering;
Fig. 6 is the testing result on the earth road surface of embodiment 2, and left side figure is the road segmentation result based on the maximum entropy threshold method, and right figure is the road segmentation result of standard Otsu threshold method;
Fig. 7 is the testing result of the cement pavement of embodiment 3, and left side figure is the original image of cement road, and right figure is the road segmentation result of edge method for registering;
Fig. 8 is the testing result of the cement pavement of embodiment 3, and left side figure is the road segmentation result based on the maximum entropy threshold method, and right figure is the road segmentation result of standard Otsu threshold method.
Embodiment
Below in conjunction with specific embodiment the present invention is done further explanation.
In order to verify the validity of this detection method, under dissimilar destructuring road scene, done the Road Detection experiment, and compared with standard Otsu dual threshold method and maximum entropy dual threshold method.All experiments of system all are experiment tests that the off-line video flowing under several sections that gather on the intelligent vehicle of reality different scenes is done, and intelligent vehicle is less than 100ms to the time requirement of entire image processing system.Hardware environment is that the computing machine of USB2.0 interface, dominant frequency 2.4G, internal memory 2G and the resolution of image are set to 320 * 240.
Embodiment 1
Based on the destructuring Road Detection method of adaptive edge registration, be the method for registering at Otsu edge and weighting Canny edge.Left side figure among Fig. 1 is the sand-gravel surface original image that collects on the intelligent vehicle, and right figure is the Canny edge image, as the scale of weighing image segmentation.Concrete grammar is:
(1) weighting Canny rim detection
At the dual threshold detection-phase, record is identified as the gradient magnitude of edge pixel, unites the irregular border that characterizes the destructuring road with length, angle, distance and the average amplitude intensity of broken line;
At the dual threshold detection-phase, the marginal point that communicates with each other is unified numbering, and after numbering is accomplished, utilize least square method to carry out fitting a straight line, obtain the axis of edge point set with identical numbering;
Utilize formula (1) to any point in the edge point set
iCompose weights, wherein S is always counting of edge point set;
Be line segment
The number of last pixel,
L Th Be line of shortest length segment length threshold value,
Be scale factor, be used for regulating
Influence degree to weights;
(2) the threshold value optimization of Otsu
All Canny marginal points in the image calculate the weighting grey level histogram
, wherein
For gray scale does
iThe number of marginal point,
Be pixel
Weights;
In the Otsu threshold value
T Best
Near search for, choose the weighting grey level histogram
Middle distance
T Best Nearest maximum point is as the new threshold value of Otsu algorithm; The decision criteria of maximum point does
, wherein
For nearest maximum point arrives
T Best Distance;
(3) the weights revaluation at Canny edge
Adopt the weights revaluation of formula (2) to the Canny edge;
Formula (2) is to being that corresponding point on the diagonal of a matrix of 2r*2r at center are judged with the Canny edge; When occurring on the diagonal line differing from the value of neighbor pixel; This marginal point Canny edge will be able to keep, and wherein
is the xor operation symbol.
The result as shown in Figures 2 and 3, the left side figure among Fig. 2 is the segmentation result before this detection method registration, right figure is the segmentation result behind this detection method registration; Left figure among Fig. 3 and right figure are respectively the segmentation threshold before and after the registration.Light gray is the grey level histogram of original image among the figure, and black is edge histogram, and through the edge registration, the registration of Otsu edge and road boundary increases, and makes the Region Segmentation threshold value more and more accurate; Left figure among Fig. 4 and right figure are respectively the road segmentation result based on maximum entropy threshold method and standard Otsu threshold method; Can find out that the maximum entropy threshold method is divided into non-road area with the segment path zone errors, and standard Otsu threshold value rule and detection method performance of the present invention are approaching, can be good at cutting apart road area and non-road area, and have higher degree of accuracy.Red lines in split image are represented filtered weighted edge.
Embodiment 2
With reference to the detection method of embodiment 1, wherein, the left side figure of Fig. 5 is the original image of the muddy and ruthed lane that collects on the intelligent vehicle, and the right side figure of Fig. 5 is the road segmentation result of edge of the present invention method for registering; The left side figure of Fig. 6 is the road segmentation result based on the maximum entropy threshold method; The right side figure of Fig. 6 is the road segmentation result of standard Otsu threshold method; Obviously, the two kinds of methods in back receive the influence of road shade to be divided into two parts to road, and detection method of the present invention can well split road.
Embodiment 3
With reference to the detection method of embodiment 1, wherein, the left side figure of Fig. 7 is the cement road original image that is distributed with shadow that collects on the intelligent vehicle, and the right figure of Fig. 7 is the road segmentation result figure of edge of the present invention method for registering; Fig. 8 left side figure is the road segmentation result of maximum entropy threshold method, and the right figure of Fig. 8 is the road segmentation result of standard Otsu threshold method.The maximum entropy threshold method receives the influence of shade bigger, and standard Otsu threshold method receives the influence of shade smaller, and edge of the present invention method for registering receives the influence of shadow minimum.
Claims (2)
1. the destructuring Road Detection method based on the adaptive edge registration is characterized in that: be the method for registering at Otsu edge and weighting Canny edge; Specifically comprise:
(1) weighting Canny rim detection
At the dual threshold detection-phase, record is identified as the gradient magnitude of edge pixel, unites the irregular border that characterizes the destructuring road with length, angle, distance and the average amplitude intensity of broken line;
At the dual threshold detection-phase, the marginal point that communicates with each other is unified numbering, and after numbering is accomplished, utilize least square method to carry out fitting a straight line, obtain the axis of edge point set with identical numbering;
Utilize formula (1) to any point in the edge point set
iCompose weights, wherein S is always counting of edge point set;
Be line segment
The number of last pixel,
L Th Be line of shortest length segment length threshold value,
Be scale factor, span is 0 ~ 1, is used for regulating
Influence degree to weights;
(2) the threshold value optimization of Otsu
All Canny marginal points in the image calculate the weighting grey level histogram
, wherein
For gray scale does
iThe number of marginal point,
Be pixel
Weights;
In the Otsu threshold value
T Best
Near search for, choose the weighting grey level histogram
Middle distance
T Best Nearest maximum point is as the new threshold value of Otsu algorithm; The decision criteria of maximum point does
, wherein
For nearest maximum point arrives
T Best Distance;
(3) the weights revaluation at Canny edge
Adopt the weights revaluation of formula (2) to the Canny edge;
is weighting Canny edge image in the formula; Locating to get at 0 o'clock at
is non-edge pixel, and value was an edge pixel more than or equal to 1 o'clock;
is the Otsu image after optimizing; Getting at 1 o'clock is road area, gets 0 or be non-road area at-1 o'clock;
Formula (2) is to being that corresponding point on the diagonal of a matrix of 2r*2r at center are judged with the Canny edge; When occurring on the diagonal line differing from the value of neighbor pixel; This marginal point Canny edge will be able to keep, and wherein
is the xor operation symbol.
2. the destructuring Road Detection method based on the adaptive edge registration according to claim 1, it is characterized in that: described destructuring road is: road area and non-road area will have the difference on the gray scale; No matter be the muddy and ruthed lane or the gravel road of cement road, asphalt road or country, most of zone of road area all has the characteristics of gradation uniformity and gradually changeable; Specific region in the image can be regarded as road area.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110341479 CN102436644B (en) | 2011-11-02 | 2011-11-02 | Unstructured road detection method based on adaptive edge registration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110341479 CN102436644B (en) | 2011-11-02 | 2011-11-02 | Unstructured road detection method based on adaptive edge registration |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102436644A true CN102436644A (en) | 2012-05-02 |
| CN102436644B CN102436644B (en) | 2013-10-16 |
Family
ID=45984689
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201110341479 Active CN102436644B (en) | 2011-11-02 | 2011-11-02 | Unstructured road detection method based on adaptive edge registration |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102436644B (en) |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102682292A (en) * | 2012-05-10 | 2012-09-19 | 清华大学 | Method based on monocular vision for detecting and roughly positioning edge of road |
| CN102842039A (en) * | 2012-07-11 | 2012-12-26 | 河海大学 | Road image detection method based on Sobel operator |
| CN103310006A (en) * | 2013-06-28 | 2013-09-18 | 电子科技大学 | ROI extraction method in auxiliary vehicle driving system |
| CN103413144A (en) * | 2013-07-29 | 2013-11-27 | 西北工业大学 | Airport detection and recognition method based on local global feature joint decision |
| CN103473763A (en) * | 2013-08-31 | 2013-12-25 | 哈尔滨理工大学 | Road edge detection method based on heuristic probability Hough transformation |
| CN103577828A (en) * | 2013-11-22 | 2014-02-12 | 中国科学院自动化研究所 | Road detection method based on edge feature |
| CN103714538A (en) * | 2013-12-20 | 2014-04-09 | 中联重科股份有限公司 | Road edge detection method, device and vehicle |
| CN104599271A (en) * | 2015-01-20 | 2015-05-06 | 中国科学院半导体研究所 | CIE Lab color space based gray threshold segmentation method |
| CN105069411A (en) * | 2015-07-24 | 2015-11-18 | 深圳市佳信捷技术股份有限公司 | Road recognition method and device |
| CN106384358A (en) * | 2016-08-30 | 2017-02-08 | 南京明辉创鑫电子科技有限公司 | Edge point self-similarity based irregular image recognition method |
| CN106444765A (en) * | 2016-10-21 | 2017-02-22 | 广东工业大学 | AGV (automatic guided vehicle), AGV navigation method based on vision and AGV navigation system based on vision |
| CN107229938A (en) * | 2017-08-05 | 2017-10-03 | 南京云计趟信息技术有限公司 | A kind of vehicle tire lug mud identifying system and method based on camera device |
| CN108335404A (en) * | 2018-02-07 | 2018-07-27 | 深圳怡化电脑股份有限公司 | Edge fitting method and money-checking equipment |
| CN108470343A (en) * | 2017-02-23 | 2018-08-31 | 南宁市富久信息技术有限公司 | A kind of improved method for detecting image edge |
| CN108961353A (en) * | 2017-05-19 | 2018-12-07 | 上海蔚来汽车有限公司 | The building of road model |
| CN109995860A (en) * | 2019-03-29 | 2019-07-09 | 南京邮电大学 | Deep learning task allocation algorithms based on edge calculations in a kind of VANET |
| CN110570411A (en) * | 2019-09-05 | 2019-12-13 | 中国科学院长春光学精密机械与物理研究所 | mura detection method and device based on coefficient of variation |
| CN110807771A (en) * | 2019-10-31 | 2020-02-18 | 长安大学 | Defect detection method for road deceleration strip |
| CN111080665A (en) * | 2019-12-31 | 2020-04-28 | 歌尔股份有限公司 | Image frame identification method, device and equipment and computer storage medium |
| CN112505724A (en) * | 2020-11-24 | 2021-03-16 | 上海交通大学 | Road negative obstacle detection method and system |
| CN113450292A (en) * | 2021-06-17 | 2021-09-28 | 重庆理工大学 | High-precision visual positioning method for PCBA parts |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001076147A (en) * | 1999-09-03 | 2001-03-23 | Nec Corp | System and method for detecting road white line and recording medium with program for road white line detection recorded thereon |
| US20040042638A1 (en) * | 2002-08-27 | 2004-03-04 | Clarion Co., Ltd. | Method for detecting position of lane marker, apparatus for detecting position of lane marker and alarm apparatus for lane deviation |
| CN101620732A (en) * | 2009-07-17 | 2010-01-06 | 南京航空航天大学 | Visual detection method of road driving line |
| CN102156977A (en) * | 2010-12-22 | 2011-08-17 | 浙江大学 | Vision-based road detection method |
-
2011
- 2011-11-02 CN CN 201110341479 patent/CN102436644B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001076147A (en) * | 1999-09-03 | 2001-03-23 | Nec Corp | System and method for detecting road white line and recording medium with program for road white line detection recorded thereon |
| US20040042638A1 (en) * | 2002-08-27 | 2004-03-04 | Clarion Co., Ltd. | Method for detecting position of lane marker, apparatus for detecting position of lane marker and alarm apparatus for lane deviation |
| CN101620732A (en) * | 2009-07-17 | 2010-01-06 | 南京航空航天大学 | Visual detection method of road driving line |
| CN102156977A (en) * | 2010-12-22 | 2011-08-17 | 浙江大学 | Vision-based road detection method |
Non-Patent Citations (1)
| Title |
|---|
| 王燕清 等: "基于单目视觉的非结构化道路检测与跟踪", 《哈尔滨工程大学学报》 * |
Cited By (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102682292A (en) * | 2012-05-10 | 2012-09-19 | 清华大学 | Method based on monocular vision for detecting and roughly positioning edge of road |
| CN102842039B (en) * | 2012-07-11 | 2015-06-24 | 河海大学 | Road image detection method based on Sobel operator |
| CN102842039A (en) * | 2012-07-11 | 2012-12-26 | 河海大学 | Road image detection method based on Sobel operator |
| CN103310006A (en) * | 2013-06-28 | 2013-09-18 | 电子科技大学 | ROI extraction method in auxiliary vehicle driving system |
| CN103413144A (en) * | 2013-07-29 | 2013-11-27 | 西北工业大学 | Airport detection and recognition method based on local global feature joint decision |
| CN103473763A (en) * | 2013-08-31 | 2013-12-25 | 哈尔滨理工大学 | Road edge detection method based on heuristic probability Hough transformation |
| CN103473763B (en) * | 2013-08-31 | 2017-06-20 | 哈尔滨理工大学 | Road edge detection method based on heuristic Probabilistic Hough Transform |
| CN103577828A (en) * | 2013-11-22 | 2014-02-12 | 中国科学院自动化研究所 | Road detection method based on edge feature |
| CN103714538A (en) * | 2013-12-20 | 2014-04-09 | 中联重科股份有限公司 | Road edge detection method, device and vehicle |
| CN104599271A (en) * | 2015-01-20 | 2015-05-06 | 中国科学院半导体研究所 | CIE Lab color space based gray threshold segmentation method |
| CN104599271B (en) * | 2015-01-20 | 2017-04-12 | 中国科学院半导体研究所 | CIE Lab color space based gray threshold segmentation method |
| CN105069411A (en) * | 2015-07-24 | 2015-11-18 | 深圳市佳信捷技术股份有限公司 | Road recognition method and device |
| CN105069411B (en) * | 2015-07-24 | 2019-03-29 | 深圳市佳信捷技术股份有限公司 | Roads recognition method and device |
| CN106384358A (en) * | 2016-08-30 | 2017-02-08 | 南京明辉创鑫电子科技有限公司 | Edge point self-similarity based irregular image recognition method |
| CN106384358B (en) * | 2016-08-30 | 2019-03-29 | 南京明辉创鑫电子科技有限公司 | The recognition methods of irregular image based on marginal point self-similarity |
| CN106444765A (en) * | 2016-10-21 | 2017-02-22 | 广东工业大学 | AGV (automatic guided vehicle), AGV navigation method based on vision and AGV navigation system based on vision |
| CN106444765B (en) * | 2016-10-21 | 2019-07-09 | 广东工业大学 | A kind of AGV air navigation aid of view-based access control model |
| CN108470343A (en) * | 2017-02-23 | 2018-08-31 | 南宁市富久信息技术有限公司 | A kind of improved method for detecting image edge |
| CN108961353A (en) * | 2017-05-19 | 2018-12-07 | 上海蔚来汽车有限公司 | The building of road model |
| CN108961353B (en) * | 2017-05-19 | 2023-12-08 | 上海蔚来汽车有限公司 | Construction of road model |
| CN107229938A (en) * | 2017-08-05 | 2017-10-03 | 南京云计趟信息技术有限公司 | A kind of vehicle tire lug mud identifying system and method based on camera device |
| CN108335404A (en) * | 2018-02-07 | 2018-07-27 | 深圳怡化电脑股份有限公司 | Edge fitting method and money-checking equipment |
| CN109995860B (en) * | 2019-03-29 | 2022-03-04 | 南京邮电大学 | Deep learning task allocation algorithm based on edge calculation in VANET |
| CN109995860A (en) * | 2019-03-29 | 2019-07-09 | 南京邮电大学 | Deep learning task allocation algorithms based on edge calculations in a kind of VANET |
| CN110570411A (en) * | 2019-09-05 | 2019-12-13 | 中国科学院长春光学精密机械与物理研究所 | mura detection method and device based on coefficient of variation |
| CN110807771A (en) * | 2019-10-31 | 2020-02-18 | 长安大学 | Defect detection method for road deceleration strip |
| CN110807771B (en) * | 2019-10-31 | 2022-03-22 | 长安大学 | Defect detection method for road deceleration strip |
| CN111080665A (en) * | 2019-12-31 | 2020-04-28 | 歌尔股份有限公司 | Image frame identification method, device and equipment and computer storage medium |
| CN111080665B (en) * | 2019-12-31 | 2023-06-09 | 歌尔光学科技有限公司 | Image frame recognition method, device, equipment and computer storage medium |
| CN112505724A (en) * | 2020-11-24 | 2021-03-16 | 上海交通大学 | Road negative obstacle detection method and system |
| CN113450292A (en) * | 2021-06-17 | 2021-09-28 | 重庆理工大学 | High-precision visual positioning method for PCBA parts |
| CN113450292B (en) * | 2021-06-17 | 2022-08-16 | 重庆理工大学 | High-precision visual positioning method for PCBA parts |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102436644B (en) | 2013-10-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102436644B (en) | Unstructured road detection method based on adaptive edge registration | |
| CN105718860B (en) | Localization method and system based on driving safety map and binocular Traffic Sign Recognition | |
| Niu et al. | Robust lane detection using two-stage feature extraction with curve fitting | |
| Li et al. | Road detection algorithm for autonomous navigation systems based on dark channel prior and vanishing point in complex road scenes | |
| CN105160309B (en) | Three lanes detection method based on morphological image segmentation and region growing | |
| Huang et al. | On-board vision system for lane recognition and front-vehicle detection to enhance driver's awareness | |
| CN103383733B (en) | A kind of track based on half machine learning video detecting method | |
| US8731244B2 (en) | Systems and methods for improving image recognition | |
| CN103714538B (en) | road edge detection method, device and vehicle | |
| Lian et al. | DeepWindow: Sliding window based on deep learning for road extraction from remote sensing images | |
| CN107958183A (en) | A kind of city road network information automation extraction method of high-resolution remote sensing image | |
| CN109271928B (en) | Road network updating method based on vector road network fusion and remote sensing image verification | |
| CN108280450A (en) | A kind of express highway pavement detection method based on lane line | |
| Zhang et al. | Semi-automatic road tracking by template matching and distance transformation in urban areas | |
| CN102982304B (en) | Utilize polarized light image to detect the method and system of vehicle location | |
| CN108845569A (en) | Generate semi-automatic cloud method of the horizontal bend lane of three-dimensional high-definition mileage chart | |
| CN106842231A (en) | A kind of road edge identification and tracking | |
| CN101916373B (en) | Road semiautomatic extraction method based on wavelet detection and ridge line tracking | |
| CN105426864A (en) | Multiple lane line detecting method based on isometric peripheral point matching | |
| CN107491756B (en) | Lane direction information recognition methods based on traffic sign and surface mark | |
| CN108171695A (en) | A kind of express highway pavement detection method based on image procossing | |
| CN103473763A (en) | Road edge detection method based on heuristic probability Hough transformation | |
| CN102354457A (en) | General Hough transformation-based method for detecting position of traffic signal lamp | |
| CN105117726A (en) | License plate positioning method based on multi-feature area accumulation | |
| LU504084B1 (en) | A non-rigid registration method of vehicle-borne LiDAR point clouds by fusing multiple road features |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20231211 Address after: 210000 10 Ma Qun Road, Qixia District, Nanjing, Jiangsu. Patentee after: JIANGSU INTELLITRAINS CO.,LTD. Address before: No. 666 Donglin Road, Qilin Technology Innovation Park, Nanjing City, Jiangsu Province, 210000 Patentee before: Nanjing Internet of Things Research Institute Development Co.,Ltd. |
|
| TR01 | Transfer of patent right |