CN112434621A - Method for extracting characteristics of inner side edge of lane line - Google Patents
Method for extracting characteristics of inner side edge of lane line Download PDFInfo
- Publication number
- CN112434621A CN112434621A CN202011359366.XA CN202011359366A CN112434621A CN 112434621 A CN112434621 A CN 112434621A CN 202011359366 A CN202011359366 A CN 202011359366A CN 112434621 A CN112434621 A CN 112434621A
- Authority
- CN
- China
- Prior art keywords
- lane line
- side edge
- point
- edge
- equation
- 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
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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method for extracting characteristics of the inner side edge of a lane line, which is characterized in that a lane line is continuous by utilizing two frames of lane line graphs, and the lane line does not have large jump, so that the approximate position of the characteristic point of the inner side edge of the current frame lane line graph can be found out through the equation of the lane line of the previous frame, and the point with the maximum gray gradient change is searched at the approximate position; because the points on one side of the inner side edge characteristic point are all located in the interest area of the lane line, and the continuity of the inner side edge characteristic point is added, the inner side edge characteristic point can be determined, and the inner side edge of the lane line is obtained. Further, if the characteristic points of the inner side edge are not extracted, updating a lane line equation by the slope of the high-quality points, improving the positioning accuracy of the characteristic points of the inner side edge, and continuously extracting the characteristic points of the inner side edge; to reduce the noise of the updated lane line slope, a plurality of consecutive slopes may be selected to average, and the lane line equation is updated by the average.
Description
Technical Field
The invention belongs to the technical field of lane line detection, and particularly relates to a method for extracting features of an inner side edge of a lane line.
Background
With the attention of people, the advanced Assistant Driving (ADAS) is paid more and more, the lane Line Deviation Warning (LDW) and lane Line Keeping (LKA) functions are paid more and more attention, and the two functions are actions which are actually performed after the lane lines are accurately detected, and some early lane line detection schemes directly detect straight lines by using hough transform, and the scheme has a good effect on straight line detection, but the detection on the curves is not satisfactory.
At present, the lane line detection is roughly divided into two parts, the front end is used for extracting the features of the lane line, and the rear end is used for fitting a lane line equation by using the extracted features. The lane line features are further divided into lane line inboard edge features and lane line center line features, as shown in fig. 1 and 2, respectively. The central line feature of the lane line is obtained by extracting edge feature points on the left side and the right side of the lane line, and then taking the mean value of the feature point coordinates on the two sides to obtain the central line feature. Compared with the centerline feature, the difficulty of extracting the inside edge feature is higher, but the lane line trend in the actual driving process can be described more accurately, and the method is also the basis of extracting the centerline feature.
In the edge feature extraction, the most typical algorithm is the Canny edge detection algorithm, which is a multi-stage edge detection algorithm developed by john.f. cannyy in 1986, and is also considered by many people to be one of the optimal algorithms for edge detection, and the detection effect of the algorithm is shown in fig. 3. The Canny algorithm is based on three basic objectives: 1. low error rate, all edges should be found, and there is no false response; 2. the edge points should be well located, the located edge must be as close as possible to the true edge; 3. a single edge point responds, meaning that in a location where only one single edge point exists, the edger should not indicate multiple pixel edges.
As shown in fig. 4, the steps of the Canny algorithm are as follows: 1. performing Gaussian smoothing on an input image, eliminating noise and reducing error rate; 2. calculating gradient magnitude and direction to estimate edge strength and direction at each point; 3. according to the direction of the gradient, carrying out non-maximum suppression on the gradient amplitude; 4. threshold hysteresis processing, processing with dual thresholds and connecting edges.
However, the existing lane line feature extraction technology has two problems: 1. the extracted feature points cannot accurately describe the trend of the lane line, which can affect the precision of the fitted lane line and subsequent actions; 2. the feature extraction is time-consuming, mainly Canny edge operators are time-consuming, requirements are put forward for hardware platforms, some of the hardware platforms even need to be accelerated by adopting an FPGA (programmable logic device), a GPU (image processing unit) and the like, and the hardware platforms are not friendly to low-computing-power platforms.
Disclosure of Invention
The invention aims to provide a method for extracting characteristics of the inner side edge of a lane line, and solves the problems that the characteristics of the inner side edge of the lane line extracted by the existing lane line characteristic extraction technology are inaccurate, the time consumption of characteristic extraction is large, and the requirement on a characteristic extraction platform is high.
The invention provides a method for extracting characteristics of an inner side edge of a lane line, which comprises the following steps:
s1, obtaining a current frame lane line graph, extracting a region where a lane line is located from the current frame lane line graph, and performing binarization processing on the lane line graph to obtain a binarization graph, wherein the region where the lane line is located is a lane line interest region;
s2, obtaining a lane line equation based on the lane line graph of the previous frame;
s3, in the binary image, searching a point with the maximum gray gradient change at the current lane line equation, and if at least 5 points exist in a 3-by-3 area taking the point as the center and are in the interest area of the lane line, the point is a similar feature point;
s4, if other similar feature points exist near a certain similar feature point, the similar feature point is an inner side edge feature point;
s5, judging whether the characteristic points of the inner side edge are extracted completely; if so, outputting all the inner side edge characteristic points, wherein the area where the inner side edge characteristic points are located is the inner side edge of the lane line.
Further, a lane line equation is obtained according to the extracted characteristic points of the inner side edge, and the step S1 is repeated to continuously extract the inner side edge of the lane line of the next frame of lane line graph.
Further, the method comprises the steps of:
s6, if the characteristic point of the inner side edge is not extracted, judging whether the characteristic point of the inner side edge is in the interest area of the lane line; if so, the characteristic point of the inner side edge is a high-quality point;
s7, judging whether the high-quality point number exceeds a preset threshold value; if not, go to step S3; if yes, the slope of the corresponding inside edge feature point is updated with the slopes of two consecutive high-quality points, and then the lane line equation is updated, and step S3 is executed.
Further, step S7 further includes: and selecting the slopes of a plurality of continuous high-quality points, solving a mean value, and updating the lane line equation through the mean value.
Further, step S6 further includes: the same high quality point is sought around the high quality point.
Further, if the current lane line equation is a left lane line equation, searching a point with the maximum gray gradient change from left to right; and if the current lane line equation is the right lane line equation, searching the point with the maximum gray gradient change from the right side to the left side.
Further, the area where the lane line is located is extracted from the lane line graph through deep learning.
Further, based on the lane line graph of the previous frame, a lane line equation is obtained through a Canny edge operator.
The invention has the beneficial effects that: according to the method for extracting the characteristic of the inner side edge of the lane line, the continuous lane line graphs of the front frame and the rear frame are utilized, and the lane line does not jump greatly, so that the approximate position of the characteristic point of the inner side edge of the lane line graph of the current frame can be found out through the equation of the lane line of the previous frame, and the point with the maximum gray gradient change is searched at the approximate position; because the points on one side of the inner side edge characteristic point are all located in the interest area of the lane line, and the continuity of the inner side edge characteristic point is added, the inner side edge characteristic point can be determined, and the inner side edge of the lane line is obtained.
Further, if the characteristic points of the inner side edge are not extracted, updating a lane line equation by the slope of the high-quality points, improving the positioning accuracy of the characteristic points of the inner side edge, and continuously extracting the characteristic points of the inner side edge; to reduce the noise of the updated lane line slope, a plurality of consecutive slopes may be selected to average, and the lane line equation is updated by the average.
Drawings
FIG. 1 is a schematic view of the inner edge feature of the lane line of the present invention.
FIG. 2 is a schematic view of the centerline feature of the lane marking in the present invention.
FIG. 3 is a diagram illustrating the detection effect of the Canny edge detection algorithm in the present invention.
Fig. 4 is a block diagram of the Canny algorithm of the present invention.
Fig. 5 is a flowchart of the extraction of the features of the inner edge of the lane line according to the present invention.
Fig. 6 is an original drawing of a lane line according to the present invention.
FIG. 7 is a schematic diagram of a lane line interest field according to the present invention.
Fig. 8 is a diagram illustrating the binarization of lane lines in the present invention.
Fig. 9 is a schematic diagram of the extraction of the features of the inner edge of the lane line in the present invention.
Fig. 10 is a diagram showing the effect of extracting the features of the inner edge of the lane line according to the present invention.
Detailed Description
The invention will be further described with reference to the accompanying drawings in which:
in order to enable lane line detection to cope with various environments including curves and curves, reflect the actual road trend more accurately and run on a low-calculation-force platform, the invention provides a lane line inner side edge feature extraction method, as shown in fig. 5, which comprises the following steps:
s1, obtaining a current frame lane line graph as shown in fig. 6, then extracting an area where a lane line is located from the lane line graph as shown in fig. 7, and then performing binarization processing on the lane line graph to obtain a binarized graph as shown in fig. 8, where the area where the lane line is located is a lane line interest area. Wherein, the area of the lane line can be extracted from the lane line graph through deep learning. The main purpose of the invention is to accurately and quickly extract the characteristics of the inner side edge of the lane line from the binary image.
And S2, obtaining a lane line equation based on the lane line graph of the previous frame. For example, the lane line equation is obtained by the Canny edge operator, and since this step is not the invention point of the present invention, it is not described in detail.
And S3, searching a point with the maximum gray gradient change at the current lane line equation in the binary image, and if at least 5 points exist in a 3-by-3 region taking the point as the center and are in the interest region of the lane line, determining the point as a similar feature point. Because the front and rear frames of lane line graphs are continuous, the lane line does not have large jump, and therefore the approximate position of the characteristic point of the inner side edge of the lane line in the current frame of lane line graph can be determined through the lane line equation of the previous frame. Because the characteristic point of the inner side edge is positioned at the edge of the lane line, the points on one side are all positioned in the interest area of the lane line, and similar characteristic points can be screened from the points with the maximum gray gradient change according to the characteristic points. Preferably, if the current lane line equation is a left lane line equation, searching a point with the maximum gray gradient change from left to right; and if the current lane line equation is the right lane line equation, searching the point with the maximum gray gradient change from the right side to the left side.
And S4, if other similar feature points exist near a certain similar feature point, the similar feature point is an inner edge feature point. Since the inner edge feature points of the lane line are continuous and non-isolated, other similar feature points exist in the vicinity thereof, and thus a desired inner edge feature point can be found from the similar feature points.
S5, judging whether the characteristic points of the inner side edge are extracted completely; if so, outputting all the inner side edge characteristic points, wherein the area where the inner side edge characteristic points are located is the inner side edge of the lane line.
At this time, according to the characteristic points of the inner side edge extracted in the above steps, the lane line equation of the current frame can be obtained, and the inner side edge of the lane line of the next frame of lane line graph can be continuously extracted by repeating the above steps.
Further, the method comprises the steps of:
s6, if the characteristic point of the inner side edge is not extracted, judging whether the characteristic point of the inner side edge is in the interest area of the lane line; if so, the characteristic point of the inner side edge is a high-quality point; a high quality point set can be established; and searching the same high-quality point near the high-quality point, such as the next row and the same column, and so on until the high-quality point cannot be found, and then adding the obtained high-quality points into the high-quality point set.
S7, judging whether the high-quality point number exceeds a preset threshold value; if not, go to step S3; if yes, the slope of the corresponding inside edge feature point is updated with the slopes of two consecutive high-quality points, and then the lane line equation is updated, and step S3 is executed.
Further, step S7 further includes: in order to reduce the noise of the lane line slope, the slopes of a plurality of consecutive high-quality points may be selected, and an average value may be obtained by which the lane line equation is updated.
FIG. 10 is a graph of the effect of extraction using the present invention, and the following table shows that feature extraction is time consuming.
TABLE 1 time consuming feature extraction
| Inside edge characteristic (ms) | Midline character (ms) | Platform | |
| Time consuming | 0.5 | 1.7 | Intel i7-8750,2.21GHZ |
According to the method for extracting the characteristic of the inner side edge of the lane line, the continuous lane line graphs of the front frame and the rear frame are utilized, and the lane line does not jump greatly, so that the approximate position of the characteristic point of the inner side edge of the lane line graph of the current frame can be found out through the equation of the lane line of the previous frame, and the point with the maximum gray gradient change is searched at the approximate position; because the points on one side of the inner side edge characteristic point are all located in the interest area of the lane line, and the continuity of the inner side edge characteristic point is added, the inner side edge characteristic point can be determined, and the inner side edge of the lane line is obtained. Further, if the characteristic points of the inner side edge are not extracted, updating a lane line equation by the slope of the high-quality points, improving the positioning accuracy of the characteristic points of the inner side edge, and continuously extracting the characteristic points of the inner side edge; to reduce the noise of the updated lane line slope, a plurality of consecutive slopes may be selected to average, and the lane line equation is updated by the average.
It will be understood by those skilled in the art that the foregoing is merely a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included within the scope of the present invention.
Claims (8)
1. A method for extracting characteristics of an inner side edge of a lane line is characterized by comprising the following steps:
s1, obtaining a current frame lane line graph, extracting a region where a lane line is located from the current frame lane line graph, and performing binarization processing on the lane line graph to obtain a binarization graph, wherein the region where the lane line is located is a lane line interest region;
s2, obtaining a lane line equation based on the lane line graph of the previous frame;
s3, in the binary image, searching a point with the maximum gray gradient change at the current lane line equation, and if at least 5 points exist in a 3-by-3 area taking the point as the center and are in the interest area of the lane line, the point is a similar feature point;
s4, if other similar feature points exist near a certain similar feature point, the similar feature point is an inner side edge feature point;
s5, judging whether the characteristic points of the inner side edge are extracted completely; if so, outputting all the inner side edge characteristic points, wherein the area where the inner side edge characteristic points are located is the inner side edge of the lane line.
2. The method of extracting features of an inner edge of a lane line according to claim 1, wherein a lane line equation is obtained according to the extracted features of the inner edge, and step S1 is repeated to continuously extract an inner edge of the lane line of a next frame of lane line.
3. The method of extracting features of an inside edge of a lane line according to claim 1, further comprising the steps of:
s6, if the characteristic point of the inner side edge is not extracted, judging whether the characteristic point of the inner side edge is in the interest area of the lane line; if so, the characteristic point of the inner side edge is a high-quality point;
s7, judging whether the high-quality point number exceeds a preset threshold value; if not, go to step S3; if yes, the slope of the corresponding inside edge feature point is updated with the slopes of two consecutive high-quality points, and then the lane line equation is updated, and step S3 is executed.
4. The method of extracting a feature of an inner edge of a lane line according to claim 3, wherein step S7 further includes: and selecting the slopes of a plurality of continuous high-quality points, solving a mean value, and updating the lane line equation through the mean value.
5. The method of extracting a feature of an inner edge of a lane line according to claim 3, wherein step S6 further includes: the same high quality point is sought around the high quality point.
6. The method for extracting characteristics of the inner side edge of the lane line according to claim 1, wherein if the current lane line equation is a left lane line equation, a point with the maximum gray gradient change is searched from left to right; and if the current lane line equation is the right lane line equation, searching the point with the maximum gray gradient change from the right side to the left side.
7. The method of extracting a feature of an inner edge of a lane line according to claim 1, wherein a region where the lane line is located is extracted from a lane line graph by deep learning.
8. The method for extracting characteristics of the inner edge of the lane line according to claim 1, wherein a lane line equation is obtained by a Canny edge operator based on a lane line graph of a previous frame.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011359366.XA CN112434621B (en) | 2020-11-27 | 2020-11-27 | Method for extracting characteristics of inner side edge of lane line |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011359366.XA CN112434621B (en) | 2020-11-27 | 2020-11-27 | Method for extracting characteristics of inner side edge of lane line |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112434621A true CN112434621A (en) | 2021-03-02 |
| CN112434621B CN112434621B (en) | 2022-02-15 |
Family
ID=74698418
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011359366.XA Active CN112434621B (en) | 2020-11-27 | 2020-11-27 | Method for extracting characteristics of inner side edge of lane line |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112434621B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101393034A (en) * | 2007-09-21 | 2009-03-25 | 财团法人工业技术研究院 | Traffic lane prediction method and lane bias alarm system |
| CN103136341A (en) * | 2013-02-04 | 2013-06-05 | 北京航空航天大学 | Lane line reconstruction device based on Bezier curve |
| DE102012221777A1 (en) * | 2011-12-02 | 2013-06-06 | GM Global Technology Operations LLC (n. d. Ges. d. Staates Delaware) | Track tracing system for motor car, has track tracing processor converting track boundary points into cartesian vehicle co-ordinate system and adapting track boundary points to reliability-weighted model track line |
| CN104008387A (en) * | 2014-05-19 | 2014-08-27 | 山东科技大学 | Lane line detection method based on feature point piecewise linear fitting |
| CN106803061A (en) * | 2016-12-14 | 2017-06-06 | 广州大学 | A kind of simple and fast method for detecting lane lines based on dynamic area-of-interest |
| CN109409202A (en) * | 2018-09-06 | 2019-03-01 | 惠州市德赛西威汽车电子股份有限公司 | Robustness method for detecting lane lines based on dynamic area-of-interest |
-
2020
- 2020-11-27 CN CN202011359366.XA patent/CN112434621B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101393034A (en) * | 2007-09-21 | 2009-03-25 | 财团法人工业技术研究院 | Traffic lane prediction method and lane bias alarm system |
| DE102012221777A1 (en) * | 2011-12-02 | 2013-06-06 | GM Global Technology Operations LLC (n. d. Ges. d. Staates Delaware) | Track tracing system for motor car, has track tracing processor converting track boundary points into cartesian vehicle co-ordinate system and adapting track boundary points to reliability-weighted model track line |
| CN103136341A (en) * | 2013-02-04 | 2013-06-05 | 北京航空航天大学 | Lane line reconstruction device based on Bezier curve |
| CN104008387A (en) * | 2014-05-19 | 2014-08-27 | 山东科技大学 | Lane line detection method based on feature point piecewise linear fitting |
| CN106803061A (en) * | 2016-12-14 | 2017-06-06 | 广州大学 | A kind of simple and fast method for detecting lane lines based on dynamic area-of-interest |
| CN109409202A (en) * | 2018-09-06 | 2019-03-01 | 惠州市德赛西威汽车电子股份有限公司 | Robustness method for detecting lane lines based on dynamic area-of-interest |
Non-Patent Citations (1)
| Title |
|---|
| 凌诗韵等: "基于Hough变换的城市环境道路识别优化算法研究", 《机械设计与制造工程》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112434621B (en) | 2022-02-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10140529B2 (en) | Method, apparatus and device for detecting lane lines | |
| CN110084236B (en) | Image correction method and device | |
| WO2019057067A1 (en) | Image quality evaluation method and apparatus | |
| CN113409313B (en) | Wheel weld surface defect detection method based on computer vision | |
| CN110197153B (en) | Automatic wall identification method in house type graph | |
| US9275030B1 (en) | Horizontal and vertical line detection and removal for document images | |
| CN101170641A (en) | A method for image edge detection based on threshold sectioning | |
| CN114399522A (en) | High-low threshold-based Canny operator edge detection method | |
| CN113674206B (en) | Extraction method suitable for characteristic parameters of deep-melting K-TIG welding molten pool and keyhole entrance | |
| CN104063866A (en) | Method for detecting granularity in ore transmission process | |
| CN105261021A (en) | Method and apparatus of removing foreground detection result shadows | |
| Somasundaram et al. | Separation of overlapped chromosomes and pairing of similar chromosomes for karyotyping analysis | |
| Madian et al. | Analysis of human chromosome classification using centromere position | |
| CN111127498A (en) | Canny edge detection method based on edge self-growth | |
| CN114881965A (en) | Wood board joint detection method based on artificial intelligence and image processing | |
| CN110163894B (en) | Sub-pixel level target tracking method based on feature matching | |
| CN117094975A (en) | Method and device for detecting surface defects of steel and electronic equipment | |
| CN103455985A (en) | Road crack enhancement method based on Hessian structural analysis | |
| CN111310754A (en) | Method for segmenting license plate characters | |
| CN112434621B (en) | Method for extracting characteristics of inner side edge of lane line | |
| Saini | Document image binarization techniques, developments and related issues: a review | |
| CN104268845A (en) | Self-adaptive double local reinforcement method of extreme-value temperature difference short wave infrared image | |
| CN105760876A (en) | Vehicle logo location method based on radiator grid background ablation | |
| CN107832732B (en) | Lane line detection method based on treble traversal | |
| CN110705568B (en) | Optimization method for image feature point extraction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |