Lane boundary detector

• US 7,937,196 B2
• Filed: 03/09/2005
• Issued: 05/03/2011
• Est. Priority Date: 03/12/2004
• Status: Active Grant

• Current Assignee: Toyota Jidosha Kabushiki Kaisha (Toyota Motor Corporation)
• Original Assignee: Toyota Jidosha Kabushiki Kaisha (Toyota Motor Corporation)
• Inventors: Nishida, Makoto, Watanabe, Akihiro
• Primary Examiner(s): Tran; Khoi
• Assistant Examiner(s): JEN, MINGJEN
• Application Number:
  US11/074,658
• Publication Number:
  US 20050209748A1
• Time in Patent Office: 2,246 Days
• Field of Search: 382/104, 701/28, 701/208, 340/436
• US Class Current: 701/23
• CPC Class Codes: G06T 2207/30256 Lane; Road marking
• CPC Class Codes: G06T 7/13 Edge detection
• CPC Class Codes: G06T 7/73 using feature-based methods
• CPC Class Codes: G06V 10/44 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
• 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

Abstract

A lane boundary detector includes an edge point extracting unit that extracts a plurality of edge points of a lane sign from the input image, and an edge line extracting unit that extracts a plurality of edge lines of the lane sign from the edge points to formulate a position and a direction of each of the edge lines as a straight line on the road surface, using a parameter set consisting of at least two variables. The detector further includes a branch point detecting unit that determines whether a crossing point of two of the edge lines can be a branch point of the lane, based on whether a characteristic difference vector defined with the variables with respect to the two edge lines falls within a predetermined appropriate range, and a lane boundary determining unit that determines the left and the right boundary positions of the lane.

Claims

1. A lane boundary detector comprising:

a vehicle mounted camera that obtains an input image to detect, in real time while a vehicle is running, a right and a left boundary position of a lane defined by a lane sign drawn on a road surface on which the vehicle runs;

an edge point extracting unit that extracts a plurality of edge points of the lane sign from the input image;

an edge line extracting unit that extracts a plurality of edge lines constituting a contour of the lane sign from the edge points extracted by the edge point extracting unit to formulate a position and a direction of each of the edge lines as a straight line on the road surface, using a parameter set of real numbers including at least a first variable p and a second variable q;

a branch point detecting unit that (1) verifies, based on the input image obtained by the vehicle mounted camera, whether a characteristic difference vector (p_i-p_j, q_i-q_j) defined with the first variable p and the second variable q, with respect to a first edge line i and a second edge line j, which are extracted, and a position and a direction of each of which are formulated by the edge line extracting unit falls within a predetermined appropriate range, (2) judges that a crossing point of the first edge line i and the second edge line j out of the edge lines can be a branch point of the lane when the characteristic difference vector (p_i-p_j, q_i-q_j) falls within a predetermined range, and (3) generates branch point data; and

a lane boundary determining unit that determines the left and the right boundary positions of the lane based on a determination result by the branch point detecting unit.

2. The lane boundary detector according to claim 1, wherein

when the road surface is represented by an xy-orthogonal coordinate plane where a positive direction of a y-axis represents a forward direction of the vehicle and a positive direction of an x-axis represents a rightward direction of the vehicle,

the first variable is substantially the same with an x-coordinate of the first edge line or the second edge line on a straight line that satisfies y=D in front of the vehicle where D is a predetermined distance, and

the second variable is substantially the same with an angle formed by the first edge line or the second edge line with respect to the positive direction of the y-axis.

3. The lane boundary detector according to claim 1, wherein

when the road surface is represented by a coordinate plane where a positive direction of an x-axis represents a rightward direction of the vehicle, and when the branch point detecting unit determines whether a branch point from which a branch diverts leftward is present or not,

each of the first edge line and the second edge line which are selected as determination data is an edge line that represents a set of points where value ∂

L/∂

x takes a negative minimum value, the value ∂

L/∂

x being obtained by partial differentiation of luminance L of the input image with respect to the x-coordinate.

4. The lane boundary detector according to claim 1, wherein

when the road surface is represented by a coordinate plane where a positive direction of an x-axis represents a rightward direction of the vehicle, and when the branch point detecting unit determines whether a branch point from which a branch diverts rightward is present or not,

each of the first edge line and the second edge line which are selected as determination data is an edge line that represents a set of points where value ∂

L/∂

x takes a positive maximum value, the value ∂

L/∂

x being obtained by partial differentiation of luminance L of the input image with respect to the x-coordinate.

5. The lane boundary detector according to claim 1, wherein

any one of the branch point detecting unit and the lane boundary determining unit determines whether the branch point of the lane is present or not, based on whether a characteristic difference vector (p_i-p_k, q_i-q_k) or (p_j-p_k, q_j-q_k) defined with the first variable and the second variable falls within a predetermined appropriate range with respect to at least one of the first edge line and the second edge line whose crossing point is determined likely to be the branch point of the lane and a reference edge line k(p_k, q_k) that is a predetermined object of comparison.

6. The lane boundary detector according to claim 5, wherein an edge line to be used as the reference edge line is determined based on the latest adopted boundary position of the lane.

7. The lane boundary detector according to claim 5, further comprising an image dividing unit that horizontally divides the input image into a plurality of areas, wherein an edge line that corresponds with the boundary position of the lane in a lower area than a target area in a horizontally divided image is employed as the reference edge line.

8. The lane boundary detector according to claim 1, wherein

when the branch point detecting unit determines that the crossing point is the branch point of the lane,

the lane boundary determining unit selects an edge line that is located within a range of appropriate distance from a center line of a lane on which the vehicle is currently running and that is located closest to the center line among the edge lines, as one of the left boundary position and the right boundary position with respect to the vehicle which are to be sought in the end.

9. A lane boundary detector comprising:

a vehicle mounted camera that obtains an input image to detect, in real time while a vehicle is running, a right and a left boundary position of a lane defined by a lane sign drawn on a road surface on which a vehicle runs;

an edge point extracting unit that extracts a plurality of edge points of the lane sign from the input image;

an edge line extracting unit that extracts a plurality of edge lines constituting a contour of the lane sign from the edge points extracted by the edge point extracting unit to formulate a position and a direction of each of the edge lines as a straight line on the road surface, using a parameter set of real numbers including at least a first variable p and a second variable q; and

a validity verifying unit that (1) verifies, based on the input image obtained by the vehicle mounted camera, whether a characteristic difference vector (p_i-p_k, q_i-q_k) defined with the first variable p and the second variable q falls within a predetermined appropriate range or not with respect to a first edge line i, which is extracted, and a position and a direction of each of which are formulated by the edge line extracting unit and a reference edge line k(p_k, q_k) that is a predetermined object of comparison, (2) judges a validity of the first edge line i out of the edge lines as the boundary position of the lane when the characteristic difference vector (p_i-p_k, q_i-q_k) falls within a predetermined range, (3) judges that a crossing point of the first edge line i and the reference edge line k(p_k, q_k) out of the edge lines can be a branch point of the lane, and (4) generates branch point data.

10. The lane boundary detector according to claim 9, wherein an edge line to be used as the reference edge line is determined based on the latest adopted boundary position of the lane.

11. The lane boundary detector according to claim 9, further comprising an image dividing unit that horizontally divides the input image into a plurality of areas, wherein an edge line that corresponds with the boundary position of the lane in a lower area than a target area in a horizontally divided image is employed as the reference edge line.