Vehicle detection apparatus

• US 8,976,999 B2
• Filed: 02/03/2012
• Issued: 03/10/2015
• Est. Priority Date: 02/16/2011
• Status: Active Grant

• Current Assignee: Subaru Corp.
• Original Assignee: Fuji Heavy Industries Limited (Subaru Corp.)
• Inventors: Saito, Toru
• Primary Examiner(s): LE, BRIAN Q
• Application Number:
  US13/365,321
• Publication Number:
  US 20120207348A1
• Time in Patent Office: 1,131 Days
• Field of Search: 382/103, 382/104, 382/106, 382/107, 382/108, 382/199
• US Class Current: 382/103
• CPC Class Codes: G06V 10/25 Determination of region of interest [ROI] or a volume of interest [VOI]
• CPC Class Codes: G06V 10/75 Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
• CPC Class Codes: G06V 20/584 of vehicle lights or traffic lights
• CPC Class Codes: G08G 1/04 using optical or ultrasonic detectors

Abstract

A vehicle detection apparatus includes a lamp candidate extraction unit that extracts a pixel region that may correspond to a tail lamp of a vehicle from pixel regions that an integration processing unit creates by extracting and integrating pixels of an image as a lamp candidate and a grouping unit that regroups groups containing the lamp candidate of the groups generated by grouping position data detected by a position detection unit and then regroups all groups. In the regrouping processing, a threshold used for regrouping groups containing the lamp candidate is set easier for regrouping than a threshold used for subsequently regrouping all groups.

Claims

1. A vehicle detection apparatus, comprising:

a processor configured to perform the following units;

a position detection unit that detects position data including distance information in a real space of each pixel in an image captured by an imaging unit;

a lamp candidate extraction unit that, based on a value of the pixel, extracts a pixel region in the image, as a lamp candidate, the pixel region being likely to correspond to a tail lamp of a vehicle;

a grouping unit that perfoms grouping to make points close to each other in the real space a group based on the distance information in the real space detected by the position detection unit; and

a regrouping unit that perfoms regrouping to make groups which are close within a predetermined distance from each other in the real space a group, wherein

the regrouping unit perfoms the regrouping for the groups which contain the lamp candidate in preference to the groups which do not contain the lamp candidate.

2. The vehicle detection apparatus according to claim 1, wherein, if as a result of the regrouping, the group contains the lamp candidate extracted by the lamp candidate extraction unit, the grouping unit identifies the group as a group corresponding to a rear portion of the vehicle.

3. The vehicle detection apparatus according to claim 1, further comprising a lamp pair candidate extraction unit that extracts a combination of the lamp candidates that correspond to the tail lamps including the left and right tail lamps of the vehicle as a lamp pair candidate from the lamp candidates extracted by the lamp candidate extraction unit based on the distance information in the real space detected by the position detection unit, wherein

if the group contains the lamp pair candidate extracted by the lamp pair candidate extraction unit, the grouping unit forcibly separates the group into the group that contains the lamp pair candidate and the group that does not contain the lamp candidates so as to perform regrouping as the separate groups.

4. The vehicle detection apparatus according to claim 3, wherein

the lamp pair candidate extraction unit selects any of two lamp candidates from among the lamp candidates extracted by the lamp candidate extraction unit,

extracts a combination of the two lamp candidates as the lamp pair candidate if a ratio of the number of pixels of one lamp candidate to the number of pixels of the other lamp candidate is within a predetermined range, if a difference the number of pixels in the vertical direction in the image between the two lamp candidates is within a predetermined range, if a distance and a height of the one lamp candidate in a real space are within predetermined ranges from a distance and a height of the other lamp candidates in the real space, or if an interval between the two lamp candidates in a left and right direction in the real space is within a width corresponding to one vehicle, and

performs the extraction processing for all combinations of the lamp candidates extracted by the lamp candidate extraction unit.

5. The vehicle detection apparatus according to claim 3, wherein if there is another lamp candidate between and above left and right lamp candidates which are the extracted lamp pair candidate and positioned above the left and right lamp candidates in the image, the lamp pair candidate extraction unit adds the another lamp candidate to the lamp pair candidate as a high-mounted stop lamp.

6. The vehicle detection apparatus according to claim 5, wherein if the lamp candidate added as the high-mounted stop lamp belongs to another lamp pair candidate other than the lamp pair candidate to which the lamp candidate has been added, the lamp pair candidate extraction unit excludes the added lamp candidate from the lamp pair candidate to which the lamp candidate has been added and also cancels designation of the another lamp pair candidate to which the lamp candidate whose addition is excluded belongs as the lamp pair candidate.

7. The vehicle detection apparatus according to claim 1, wherein the lamp candidate extraction unit extracts the pixels having a brightness equal to or higher than predetermined brightness in the image and, if the extracted pixels are adjacent to each other in the image, integrates the pixels as a same pixel region, calculates the height of the pixel region from a road surface based on the distance information in the real space, and extracts the pixel regions whose height from the road surface is within a predetermined range as the lamp candidates from among the pixel regions.

8. The vehicle detection apparatus according to claim 1, further comprising:

a distance image creation unit that creates a distance image by allocating the distance information in the real space detected by the position detection unit to a corresponding pixel of the image captured by the imaging unit, wherein

the grouping unit divides the distance image created by the distance image creation unit into a plurality of segments in a thin rectangular shape extending in a vertical direction with a predetermined pixel width,

creates a histogram for each of the segments,

calculates a representative distance of the segment by polling the distance information in the real space allocated to each pixel in the lamp candidate for the histogram if the lamp candidate extracted by the lamp candidate extraction unit is present in the segment,

calculates a representative distance of the segment by polling the distance information in the real space allocated to each pixel in the segment for the histogram if there is no lamp candidate extracted by the lamp candidate extraction unit in the segment, and

selects the representative distances for the grouping.

9. The vehicle detection apparatus according to claim 8, wherein in the case where the representative distance of the segment is calculated by polling the distance information in the real space allocated to each pixel in the segment for the histogram when there is not lamp candidate extracted by the lamp candidate extraction unit in the segment, the grouping unit invalidates the calculated representative distance for the segment in which a frequency of a class of the histogram corresponding to the calculated representative distance is smaller than a predetermined value and performs the grouping by assuming that there is no representative distance in the segment.

10. A vehicle detection apparatus, comprising:

a processor configured to perform the following units;

a position detection unit that detects position data including distance information in the real space of each pixel in an image captured by an imaging unit;

a lamp candidate extraction unit that, based on the value of the pixel, extracts a pixel region on the image, as a lamp candidate, the pixel region being likely to correspond to a tail lamp of a vehicle;

a grouping unit that perfoms grouping to make points closed to each other in the real space a group based on the distance information in the real space detected by the position detection unit; and

a regrouping unit that perfoms regrouping to make the groups which are close within a threshold distance from each other in the real space a group, wherein

the regrouping unit sets the threshold distance for peforming the regrouping for the groups which contain the lamp candidate and the threshold distance for peforming the regrouping for the groups which do not contain the lamp candidate, such that the regrouping is performed for the groups which contain the lamp candidate in preference to the groups which do not contain the lamp candidate.

11. The vehicle detection apparatus according to claim 10, wherein, if, as a result of the regrouping, the group contains the lamp candidate extracted by the lamp candidate extraction unit, the grouping unit identifies the group as a group corresponding to a rear portion of the vehicle.

12. The vehicle detection apparatus according to claim 10, further comprising a lamp pair candidate extraction unit that extracts a combination of the lamp candidates that correspond to the tail lamps including the left and right tail lamps of the vehicle as a lamp pair candidate from the lamp candidates extracted by the lamp candidate extraction unit based on the distance information in the real space detected by the position detection unit, wherein

if the group contains the lamp pair candidate extracted by the lamp pair candidate extraction unit, the grouping unit forcibly separates the group into the group that contains the lamp pair candidate and the group that does not contain the lamp candidates so as to perform regrouping as the separate groups.

13. The vehicle detection apparatus according to claim 12, wherein

the lamp pair candidate extraction unit selects any of two lamp candidates from among the lamp candidates extracted by the lamp candidate extraction unit,

extracts a combination of the two lamp candidates as the lamp pair candidate if a ratio of the number of pixels of one lamp candidate to the number of pixels of the other lamp candidate is within a predetermined range, if a difference the number of pixels in a vertical direction in the image between the two lamp candidates is within a predetermined range, if a distance and a height of the one lamp candidate in a real space are within predetermined ranges from a distance and a height of the other lamp candidates in the real space, or if an interval between the two lamp candidates in a left and right direction in the real space is within a width corresponding to one vehicle, and

performs the extraction processing for all combinations of the lamp candidates extracted by the lamp candidate extraction unit.

14. The vehicle detection apparatus according to claim 12, wherein if there is another lamp candidate between and above left and right lamp candidates which are the extracted lamp pair candidate and positioned above the left and right lamp candidates in the image, the lamp pair candidate extraction unit adds the another lamp candidate to the lamp pair candidate as a high-mounted stop lamp.

15. The vehicle detection apparatus according to claim 14, wherein if the lamp candidate added as the high-mounted stop lamp belongs to another lamp pair candidate other than the lamp pair candidate to which the lamp candidate has been added, the lamp pair candidate extraction unit excludes the added lamp candidate from the lamp pair candidate to which the lamp candidate has been added and also cancels designation of the another lamp pair candidate to which the lamp candidate whose addition is excluded belongs as the lamp pair candidate.

16. The vehicle detection apparatus according to claim 10, wherein the lamp candidate extraction unit extracts the pixels having a brightness equal to or higher than predetermined brightness in the image and, if the extracted pixels are adjacent to each other in the image, integrates the pixels as a same pixel region, calculates the height of the pixel region from a road surface based on the distance information in the real space, and extracts the pixel regions whose height from the road surface is within a predetermined range as the lamp candidates from among the pixel regions.

17. The vehicle detection apparatus according to claim 10, further comprising:

a distance image creation unit that creates a distance image by allocating the distance information in the real space detected by the position detection unit to a corresponding pixel of the image captured by the imaging unit, wherein

the grouping unit divides the distance image created by the distance image creation unit into a plurality of segments in a thin rectangular shape extending in the vertical direction with a predetermined pixel width,

creates a histogram for each of the segments,

calculates a representative distance of the segment by polling the distance information in the real space allocated to each pixel in the lamp candidate for the histogram if the lamp candidate extracted by the lamp candidate extraction unit is present in the segment,

calculates a representative distance of the segment by polling the distance information in the real space allocated to each pixel in the segment for the histogram if there is no lamp candidate extracted by the lamp candidate extraction unit in the segment, and

selects the representative distances for the grouping.

18. The vehicle detection apparatus according to claim 17, wherein in the case where the representative distance of the segment is calculated by polling the distance information in the real space allocated to each pixel in the segment for the histogram when there is not lamp candidate extracted by the lamp candidate extraction unit in the segment, the grouping unit invalidates the calculated representative distance for the segment in which a frequency of a class of the histogram corresponding to the calculated representative distance is smaller than a predetermined value and performs the grouping by assuming that there is no representative distance in the segment.

19. A vehicle detection apparatus, comprising:

a processor configured to perform the following units;

a position detection unit that detects position data including a distance of each pixel in an image captured by an imaging unit;

an integration processing unit that extracts the pixels having a brightness equal to or higher than predetermined brightness on the image and, if the extracted pixels are adjacent to each other on the image, integrates the pixels as a same pixel region;

a lamp candidate extraction unit that extracts the pixel region that correspond to a tail lamp of a vehicle from among the pixel regions integrated by the integration processing unit as a lamp candidate based on the position data detected by the position detection unit; and

a grouping unit that groups position data detected by the position detection unit, determines whether o regroup the generated groups containing the position data belonging to the lamp candidate extracted by the lamp candidate extraction unit, after regrouping the groups determined to be capable of being regrouped, determines whether to regroup all the groups, and then regroups the groups determined to be capable of being regrouped, wherein

upon regrouping processing by the grouping unit, a threshold about the position data used for the regrouping of the groups containing the position data belonging to the lamp candidate is set easier for the regrouping than a threshold about the position data for the subsequent regrouping of all the groups.