CN105912977B - Lane line detection method based on point clustering - Google Patents
Lane line detection method based on point clustering Download PDFInfo
- Publication number
- CN105912977B CN105912977B CN201610195295.1A CN201610195295A CN105912977B CN 105912977 B CN105912977 B CN 105912977B CN 201610195295 A CN201610195295 A CN 201610195295A CN 105912977 B CN105912977 B CN 105912977B
- Authority
- CN
- China
- Prior art keywords
- line
- lane
- clustering
- lane line
- detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
Abstract
The invention provides a lane line detection method based on point clustering, which applies the coordinate transformation idea used by Hough detection straight lines to our lane line detection and converts the lane line detection into clustering of point pairs; after the straight line is mapped to a k-b coordinate system from an x-y coordinate system, the central position of a clustering point in the k-b coordinate system is important, the central position comprises lane line position and slope information, and the lane lines can be effectively extracted by analyzing the information. The invention does not depend on the gray information of the image excessively, makes full use of the edge information detected by the canny edge, extracts the lane line by clustering the discrete points by mapping the straight line from the x-y coordinate system to the k-b coordinate system, and realizes the detection of the lane line.
Description
Technical Field
The invention belongs to the field of image processing and pattern recognition, and mainly relates to a lane line detection technology.
Background
The lane line detection is to accurately and quickly find out the position of the lane line in the picture from the lane picture through a proper algorithm. Therefore, the vehicle can calculate the relative position of the vehicle and the lane line through the calibration data of the camera, and the purpose of lane early warning is achieved. The performance of a lane departure system is directly influenced by the quality of a lane line detection algorithm. In practice, the existing lane line detection algorithm is easy to cause instability of threading judgment for the virtual lane line. The detected lane repeatedly jumps between the real lane line and the virtual lane line, and tracking is always wrong when the real lane line is erroneously detected.
In the field of digital images, Hough transform is an important shape object extraction technology, and particularly has a good extraction effect on straight lines, ellipses and the like. The transformation uses the voting principle in the transformation space to obtain the relevant parameter values of the best image in a specific shape. The central idea of the hough transform is as follows: the method comprises the steps of converting a certain-shaped object from one space to another space, converting a specific shape characteristic in one space into another more conveniently-calculated characteristic in another space, and then detecting the relevant parameter value of the object in any shape by using a voting principle.
In a general x-y coordinate space, any straight line can be represented by the formula y ═ k × x + b, where x and y are variables representing the x-axis and y-axis coordinates of the straight line, and k and b are scalars representing the slope and intercept of the straight line, respectively.
The idea of Hough transformation is to convert one coordinate system space to another coordinate system space, finding out the characteristic of calculating straight lines more easily. And (3) converting the straight line from the x-y coordinate system into a k-b coordinate system, namely converting the straight line from the formula y-k x + b into the formula b-x k + y. In the formula, k and b are variables respectively representing the corresponding values of the k axis and the b axis, and x and y are scalars. As shown in FIG. 1, three points in the x-y coordinate space of FIG. 1(a) correspond to 3 lines in the k-b coordinate system of FIG. 1 (b). Similarly, 3 lines in x-y coordinate space would correspond to 3 points in the k-b coordinate system.
Disclosure of Invention
The invention aims to solve the technical problem of providing a method which has higher robustness and realizes lane line detection by identifying straight lines by Hough transform.
The technical scheme adopted by the invention for solving the technical problems is that the lane line detection method based on point clustering comprises the following steps:
1) an edge detection step: intercepting the lower half part of the driving image to carry out canny edge detection to obtain an image after edge detection;
2) hough straight line detection:
carrying out Hough transform Hough line detection on the image after edge detection, and connecting the disconnected lane lines on the same line by setting a maximum line gap maxLineGap;
3) an image segmentation step:
uniformly dividing the image subjected to Hough straight line detection at equal intervals horizontally, searching straight lines in each interval, and mapping one section of searched straight line to one point of a k-b coordinate system;
4) clustering: carrying out fuzzy c-means clustering on points on a k-b coordinate system; the number of the classes is the number of the preset candidate lane lines, and the position of the central point of each class is a straight line corresponding to the candidate lane lines;
5) a lane determining step:
and selecting a left lane line with the slope being more than zero and closest to the vertical central line of the image as a correct left lane line from the candidate lane lines, and selecting a right lane line with the slope being less than zero and closest to the vertical central line of the image as a correct right lane line from the candidate lane lines.
The method applies the coordinate transformation idea used by Hough detection straight lines to the lane line detection, and converts the lane line detection into clustering of point pairs; after the straight line is mapped to a k-b coordinate system from an x-y coordinate system, the central position of a clustering point in the k-b coordinate system is important, the central position comprises lane line position and angle (slope) information, and the lane lines can be effectively extracted by analyzing the information.
The invention has the beneficial effects that a new lane line detection algorithm is provided, the edge information detected by canny edges is fully utilized without excessively depending on the gray information of images, and the lane lines are extracted by mapping straight lines from an x-y coordinate system to a k-b coordinate system to cluster discrete points, so that the detection of the lane lines is realized.
Drawings
Fig. 1 is a schematic diagram of Hough transform.
Fig. 2 is a picture obtained by the vehicle-mounted camera.
Fig. 3 is a canny edge detection diagram.
Fig. 4 is a Hough line inspection diagram.
Fig. 5 is a line expansion diagram.
Fig. 6 is a schematic diagram of a segmented image.
FIG. 7 is a graph showing the results of detection.
Detailed Description
The image of the lane line to be detected is shown in fig. 2, and the detection method is implemented in the VS2010 platform by C + + programming, and comprises the following steps:
1. canny edge detection
In order to eliminate a large amount of interference information and to keep the current lane line information, the embodiment selects the bottom 3/10 portion of the image for canny edge detection, and it can be seen from fig. 3 that the lane line edge information is kept, but some non-lane line information still remains, and it is also determined that the current vehicle is in the lane in the multi-lane case.
2. Hough line detection
And (5) carrying out Hough straight line detection on the image subjected to canny edge detection processing. The maximum straight line gap maxLineGap in the Hough detection straight line is set, so that the disconnected lane lines on the same straight line are connected, and the result is as shown in figure 5, which is important for correctly detecting the virtual lane line, and the length of the virtual lane line is expanded.
The maximum line clearance is used for judging whether two line segments with the same slope and intercept and having clearance with each other are regarded as a straight line, if the clearance is larger than the value, the line segments are regarded as two line segments, and if not, the line segments are regarded as one line segment. The specific value of the maxLineGap setting can be adjusted by those skilled in the art according to the actual virtual lane line test data. The present embodiment is set to 50 here.
3. Segmenting images
The image after Hough straight line detection is horizontally and uniformly segmented as shown in FIG. 6. A straight line is searched in each small interval and a piece of the searched straight line is mapped to a point in the k-b coordinate system, so that the k-b coordinate system will obtain a scatter diagram.
4. Clustering
And obtaining a scatter diagram of the k-b coordinate system, and carrying out fuzzy c-means clustering. The number of classes is the number of candidate lane lines, the position of the center point is the straight line corresponding to the candidate lane line, and the number of classes is set to be 4 in this embodiment.
5. Clustering problem is converted into solving of projection in vertical direction
Clustering is to cluster closely spaced points into one class, which is equivalent to finding the straight line with the maximum projection in the vertical direction, i.e. the lane line we want to detect. However, considering the characteristics of interference information and the intermittent characteristic of virtual lane lines, after the number of classes is set to be 4, the left lane and the right lane can both keep two lane candidate lines with the first two lengths, the left lane and the right lane closest to the vertical center line of the image are selected and respectively operated to obtain (y2-y1)/(x2-x1), the left result is greater than zero, and the right result is less than zero, namely the right lane is judged to be the correct lane. If the current two lane lines do not satisfy a positive-negative operation result, then a lane line with a sign opposite to that of the previous operation result is selected from the unselected lane lines as a correct lane line, and the other correct lane line is the lane line with the same sign as the previous two lane lines and closest to the center line as the current lane line, as shown in fig. 7.
Claims (4)
1. The lane line detection method based on point clustering is characterized by comprising the following steps:
1) an edge detection step: intercepting the lower half part of the driving image to carry out canny edge detection to obtain an image after edge detection;
2) hough straight line detection:
carrying out Hough transform Hough line detection on the image after edge detection, and connecting the disconnected lane lines on the same line by setting a maximum line gap maxLineGap;
3) an image segmentation step:
uniformly dividing the image subjected to Hough straight line detection at equal intervals horizontally, searching straight lines in each interval, and mapping one section of searched straight line to one point of a k-b coordinate system;
4) clustering: carrying out fuzzy c-means clustering on points on a k-b coordinate system; the number of the classes is the number of the preset candidate lane lines, and the position of the central point of each class is a straight line corresponding to the candidate lane lines;
5) a lane determining step:
and selecting a left lane line with the slope being more than zero and closest to the vertical central line of the image as a correct left lane line from the candidate lane lines, and selecting a right lane line with the slope being less than zero and closest to the vertical central line of the image as a correct right lane line from the candidate lane lines.
2. The method for detecting the lane line based on the point clustering as claimed in claim 1, wherein the step of intercepting the lower half of the driving image is specifically intercepting the driving image 3/10.
3. The point-clustering-based lane line detection method according to claim 1, wherein a maximum straight line gap maxLineGap is set to 50.
4. The method for detecting a lane line based on point clustering of claim 1, wherein the number of classes set in the clustering step is 4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610195295.1A CN105912977B (en) | 2016-03-31 | 2016-03-31 | Lane line detection method based on point clustering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610195295.1A CN105912977B (en) | 2016-03-31 | 2016-03-31 | Lane line detection method based on point clustering |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105912977A CN105912977A (en) | 2016-08-31 |
CN105912977B true CN105912977B (en) | 2021-03-30 |
Family
ID=56745258
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610195295.1A Active CN105912977B (en) | 2016-03-31 | 2016-03-31 | Lane line detection method based on point clustering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105912977B (en) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106529488A (en) * | 2016-11-18 | 2017-03-22 | 北京联合大学 | Lane line detection method based on ORB feature extraction |
JP6989766B2 (en) * | 2017-09-29 | 2022-01-12 | ミツミ電機株式会社 | Radar device and target detection method |
CN107977608B (en) * | 2017-11-20 | 2021-09-03 | 土豆数据科技集团有限公司 | Method for extracting road area of highway video image |
CN109636877B (en) * | 2018-10-31 | 2021-06-01 | 百度在线网络技术(北京)有限公司 | Lane line adjustment processing method and device and electronic equipment |
CN109829366B (en) * | 2018-12-20 | 2021-04-30 | 中国科学院自动化研究所南京人工智能芯片创新研究院 | Lane detection method, device and equipment and computer readable storage medium |
CN110345952A (en) * | 2019-07-09 | 2019-10-18 | 同济人工智能研究院(苏州)有限公司 | A kind of serializing lane line map constructing method and building system |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103839264A (en) * | 2014-02-25 | 2014-06-04 | 中国科学院自动化研究所 | Detection method of lane line |
CN104036246A (en) * | 2014-06-10 | 2014-09-10 | 电子科技大学 | Lane line positioning method based on multi-feature fusion and polymorphism mean value |
CN105069415A (en) * | 2015-07-24 | 2015-11-18 | 深圳市佳信捷技术股份有限公司 | Lane line detection method and device |
CN105160309A (en) * | 2015-08-24 | 2015-12-16 | 北京工业大学 | Three-lane detection method based on image morphological segmentation and region growing |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4988786B2 (en) * | 2009-04-09 | 2012-08-01 | 株式会社日本自動車部品総合研究所 | Boundary line recognition device |
-
2016
- 2016-03-31 CN CN201610195295.1A patent/CN105912977B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103839264A (en) * | 2014-02-25 | 2014-06-04 | 中国科学院自动化研究所 | Detection method of lane line |
CN104036246A (en) * | 2014-06-10 | 2014-09-10 | 电子科技大学 | Lane line positioning method based on multi-feature fusion and polymorphism mean value |
CN105069415A (en) * | 2015-07-24 | 2015-11-18 | 深圳市佳信捷技术股份有限公司 | Lane line detection method and device |
CN105160309A (en) * | 2015-08-24 | 2015-12-16 | 北京工业大学 | Three-lane detection method based on image morphological segmentation and region growing |
Non-Patent Citations (1)
Title |
---|
高速公路弯道场景的路面区域提取;隆迪;《中国优秀硕士学位论文全文数据库(信息科技辑)》;20160315(第03期);第I138-7410页 * |
Also Published As
Publication number | Publication date |
---|---|
CN105912977A (en) | 2016-08-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105912977B (en) | Lane line detection method based on point clustering | |
Luvizon et al. | A video-based system for vehicle speed measurement in urban roadways | |
CN107045629B (en) | Multi-lane line detection method | |
JP5699788B2 (en) | Screen area detection method and system | |
CN106204572B (en) | Road target depth estimation method based on scene depth mapping | |
US8902053B2 (en) | Method and system for lane departure warning | |
JP6062791B2 (en) | License plate character segmentation using likelihood maximization | |
CN108280450B (en) | Expressway pavement detection method based on lane lines | |
CN107424142B (en) | Weld joint identification method based on image significance detection | |
CN106778712B (en) | Multi-target detection and tracking method | |
CN111047615B (en) | Image-based straight line detection method and device and electronic equipment | |
CN106991686B (en) | A kind of level set contour tracing method based on super-pixel optical flow field | |
US20180253852A1 (en) | Method and device for locating image edge in natural background | |
CN110415296B (en) | Method for positioning rectangular electric device under shadow illumination | |
KR101742115B1 (en) | An inlier selection and redundant removal method for building recognition of multi-view images | |
JP2009163682A (en) | Image discrimination device and program | |
KR20180098945A (en) | Method and apparatus for measuring speed of vehicle by using fixed single camera | |
US9443312B2 (en) | Line parametric object estimation | |
CN112101108A (en) | Left-right-to-pass sign identification method based on pole position characteristics of graph | |
CN111695373A (en) | Zebra crossing positioning method, system, medium and device | |
Wang et al. | Self-calibration of traffic surveillance cameras based on moving vehicle appearance and 3-D vehicle modeling | |
JP5027201B2 (en) | Telop character area detection method, telop character area detection device, and telop character area detection program | |
CN103544495A (en) | Method and system for recognizing of image categories | |
JP2013080389A (en) | Vanishing point estimation method, vanishing point estimation device, and computer program | |
CN104766321A (en) | Infrared pedestrian image accurate segmentation method utilizing shortest annular path |
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 |