Lane detection system and method

• US 9,245,188 B2
• Filed: 05/05/2014
• Issued: 01/26/2016
• Est. Priority Date: 12/11/2013
• Status: Active Grant

• Current Assignee: Hanwha Aerospace Co., Ltd. (Hanwha Corporation)
• Original Assignee: Hanwha Techwin Co., Ltd. (Hanwha Corporation)
• Inventors: Han, Tea Won
• Primary Examiner(s): NAKHJAVAN, SHERVIN K
• Application Number:
  US14/269,335
• Publication Number:
  US 20150161454A1
• Time in Patent Office: 631 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: B60R 1/00 Optical viewing arrangements; Real-time viewing arrangements for drivers or passengers using optical image capturing systems, e.g. cameras or video systems specially adapted for use in or on vehicles
• CPC Class Codes: B60R 1/24 in front of the vehicle
• CPC Class Codes: B60R 1/27 providing all-round vision, e.g. using omnidirectional cameras
• CPC Class Codes: B60R 2300/804 for lane monitoring
• CPC Class Codes: G06V 20/588 Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
• CPC Class Codes: G06V 30/422 Technical drawings; Geographical maps

Abstract

Provided is a lane detection system and method. The lane detection method includes extracting feature points of a driving lane from an input road image, extracting candidate start points and candidate end points from candidate control points selected from among the feature points, selecting a left start point and a right start point of the driving lane from among the candidate start points based on information corresponding to a width of the driving lane, and selecting a left end point and a right end point of the driving lane from among the candidate end points based on a number of the candidate control points between each of the left start point and the candidate end points and the right start point and the candidate end points.

Claims

1. A lane detection method comprising:

extracting feature points of a driving lane from an input road image;

extracting candidate start points and candidate end points from candidate control points selected from among the feature points;

selecting a left start point and a right start point of the driving lane from among the candidate start points based on information corresponding to a width of the driving lane; and

selecting a left end point and a right end point of the driving lane from among the candidate end points based on a number of the candidate control points between each of the left start point and the candidate end points and the right start point and the candidate end points.

2. The lane detection method of claim 1, wherein the extracting the feature points comprises extracting the feature points from a conversion image obtained by performing a bird'"'"'s-eye view conversion of the input road image.

3. The lane detection method of claim 1, wherein the extracting the candidate start points and the candidate end points comprises:

detecting a straight-line pattern between the feature points and generating a first line based on the detected straight-line pattern;

setting a pair of the feature points which form respective end points of the first line as the candidate control points;

selecting, as a candidate start point, a point on a second line extending from the first line along a same gradient as the first line; and

selecting, as a candidate end point, an uppermost candidate control point from among candidate control points in a detection region set from a candidate control point paired with the candidate start point from among the candidate control points.

4. The lane detection method of claim 3, wherein the selecting of the candidate start point comprises selecting, as the candidate start point, a point at which a line located at a position indicating a start of a road intersects the second line.

5. The lane detection method of claim 3, wherein the detection region extends from the candidate control point paired with the candidate start point in a direction opposite to an extending direction of the second line, and comprises a same gradient as that of the second line and a width corresponding to a sum of left and right distances from the candidate control point paired with the candidate start point, and

wherein the left and right distances are perpendicular to the gradient of the second line.

6. The lane detection method of claim 1, wherein the selecting the left start point and the right start point comprises:

classifying the candidate start points into a left group and a right group;

selecting a candidate start point of the right group as the right start point, the candidate start point of the right group corresponding to a position at which a sum of probability distributions for existence of a right start point located within a distance corresponding to the width of the driving lane from a position of each candidate start point of the left group is largest; and

selecting a candidate start point of the left group as the left start point, the candidate start point of the left group corresponding to a position at which a sum of probability distributions for existence of a left start point located within a distance corresponding to the width of the lane from a position of each candidate start point of the right group is largest.

7. The lane detection method of claim 1, wherein the selecting the left end point and the right end point comprises:

selecting a left candidate end point of a third line intersecting a largest number of the candidate control points as the left end point from among third lines connecting the left start point with the respective candidate end points; and

selecting a right candidate end point of a third line intersecting a largest number of the candidate control points as the right end point from among third lines connecting the right start point with the respective candidate end points.

8. The lane detection method of claim 1, wherein the selecting the left end point and the right end point comprises:

selecting, as the left end point, a left candidate end point of a left third line having a largest number of the candidate control points existing within left and right distances from the left third line from among left third lines connecting the left start point with the respective candidate end points; and

selecting, as the right end point, a right candidate end point of a right third line having a largest number of the candidate control points existing within left and right distances from the right third line from among right third lines connecting the right start point with the respective candidate end points,

wherein the left and right distances from the left third line are perpendicular to a gradient of the left third line, and

wherein the left and rights distances from the right third line are perpendicular to a gradient of the right third line.

9. The lane detection method of claim 8, wherein the selecting the left end point and the right end point further comprises filtering non-connected candidate end points, from among the candidate end points, that are not to be connected with either of the left start point or the right start point by using a road map.

10. The lane detection method of claim 9, wherein the selecting the left end point and the right end point further comprises filtering the candidate end points based on road pattern information extracted from the road map.

11. A lane detection system comprising:

a feature point extractor configured to extract feature points of a driving lane from an input road image;

a control extractor configured to extract candidate start points and candidate end points from candidate control points selected from among the feature points; and

a lane detector configured to select a left start point and a right start point of the driving lane from among the candidate start points based on information about a width of the driving lane, and select a left end point and a right end point of the driving lane from among the candidate end points based on a number of the candidate control points between each of the left start point and the candidate end points and the right start point and the candidate end points.

12. The lane detection system of claim 11, further comprising an image converter configured to generate a conversion image obtained by performing a bird'"'"'s-eye view conversion of the input road image,

wherein the feature point extractor is further configured to extract the feature points from the conversion image.

13. The lane detection system of claim 11, wherein the control extractor comprises:

a line generator configured to detect a straight-line pattern between the feature points and generate a first line based on the detected straight-line pattern;

a candidate control point setter configured to set a pair of the feature points which form respective end points of the first line as the candidate control points;

a candidate start point selector configured to select, as a candidate start point, a point on a second line extending from the first line along a same gradient as the first line; and

a candidate end point selector configured to select, as a candidate end point, an uppermost candidate control point from among candidate control points in a detection region set from a candidate control point paired with the candidate start point from among the candidate control points.

14. The lane detection system of claim 13, wherein the candidate start point selector is further configured to select, as the candidate start point, a point at which a line located at a position indicating a start of a road intersects the second line.

15. The lane detection system of claim 13, wherein the detection region extends from the candidate control point paired with the candidate start point in a direction opposite to an extending direction of the second line, and comprises a same gradient as that of the second line and a width corresponding to a sum of left and right distances from the candidate control point paired with the candidate start point, and

wherein the left and right distances are perpendicular to the gradient of the second line.

16. The lane detection system of claim 11, wherein the lane detector comprises:

a classifier configured to classify the candidate start points into a left group and a right group; and

a start point selector configured to select a candidate start point of the right group as the right start point, the candidate start point of the right group corresponding to a position at which a sum of probability distributions for existence of a right start point located within a distance corresponding to the width of the driving lane from a position of each candidate start point of the left group is largest, and select a candidate start point of the left group as the left start point, the candidate start point of the left group corresponding to a position at which a sum of probability distributions for existence of a left start point located within a distance corresponding to the width of the driving lane from a position of each candidate start point of the right group is largest.

17. The lane detection system of claim 11, wherein the lane detector comprises an end point selector configured to select, as the left end point, a left candidate end point of a third line intersecting a largest number of the candidate control points from among third lines connecting the left start point with the respective candidate end points, and select, as the right end point, a right candidate end point of a third line intersecting a largest number of the candidate control points from among third lines connecting the right start point with the respective candidate end points.

18. The lane detection system of claim 11, wherein the lane detector comprises an end point selector configured to select, as the left end point, a left candidate end point of a left third line having a largest number of the candidate control points existing within left and right distances from the left third line from among third lines connecting the left start point with the respective candidate end points, and select, as the right end point, a right candidate end point of a right third line having a largest number of the candidate control points existing within left and right distances from the right third line from among third lines connecting the right start point with the respective candidate end points,

wherein the left and right distances from the left third line are perpendicular to a gradient of the left third line, and

wherein the left and rights distances from the right third line are perpendicular to a gradient of the right third line.

19. The lane detection system of claim 18, wherein the end point selector is further configured to filter non-connected candidate end points, from among the candidate end points, that are not to be connected with either of the left start point or the right start point by using a road map.

20. The lane detection system of claim 19, wherein the end point selector is further configured to filter the candidate end points based on road pattern information extracted from the road map.