Roadside object detection apparatus

Patent metadata

Field	Value
Publication number	US-9676330-B2
Application number	US-201214402197-A
Country	US
Kind code	B2
Filing date	Oct 26, 2012
Priority date	Jun 19, 2012
Publication date	Jun 13, 2017
Grant date	Jun 13, 2017

Abstract

A roadside object detection apparatus to detect a roadside object by analyzing images, detecting height information of an object in the images in a neighboring region having a distance less than or equal to a threshold from a vehicle, detecting a change of a height of a road surface due to the roadside object from the height information in the neighboring region, determining positions having the change of the height detected as characteristic points of the roadside object, extrapolating the characteristic points into a far-off region having a distance greater than or equal to the threshold from the vehicle, based on a road model, setting a search range for the roadside object in the far-off region based on a virtual line that is obtained by the characteristic points being extrapolated by the characteristic point extrapolation unit, and detecting the roadside object in the far-off region based on the search range.

First claim

The invention claimed is:

1. A roadside object detection apparatus to detect a roadside object using a plurality of captured images, comprising:
a processor programmed to:
analyze the captured images, and detect height information of an object in the captured images in a neighboring region, the neighboring region having a distance less than or equal to a threshold from a vehicle having the roadside object detection apparatus installed;
detect a change of a height of a road surface due to the roadside object from the height information in the neighboring region;
determine positions having the change of the height detected as characteristic points of the roadside object, define a road model based on the characteristic points, and extrapolate the road model into a far-off region, the far-off region having a distance greater than or equal to the threshold from the vehicle;
set a search range for the roadside object in the far-off region based on a virtual line that is obtained by the extrapolated road model being, and detect the roadside object in the far-off region based on the search range;
set the search range for the roadside object by translating the virtual line by predetermined distances to opposite directions in a symmetric or non-symmetric manner with respect to the virtual line; and
detect the roadside object in the far-off region by applying template matching to the search range with a template image for the neighboring region, the template image being cut off from the captured images and including the roadside object in the neighboring region.

2. The roadside object detection apparatus as claimed in claim 1 , wherein the processor is further programmed to:
detect edges in the search range, generate one or more line segments from the edges, and identify the line segments originated from the characteristic points of the roadside object in the far-off region based on at least one of the number of the line segments, and lengths of the line segments.

3. The roadside object detection apparatus as claimed in claim 1 , wherein the processor is further programmed to:
detect, based on at least one of the captured images, a lane separation mark that defines a lane in which the vehicle travels,
set the search range for the roadside object by translating the virtual line in a direction such that the translated virtual line is positioned between the virtual line and the lane separation mark that is closer to the vehicle among the lane separation marks on left and right with respect to a traveling direction, and by translating the virtual line by a predetermined distance in an opposite direction, and
detect edges in the search range, to generate one or more line segments from the edges, and identify the line segment generated from the characteristic points of the roadside object in the far-off region based on at least one of the number of the line segments, and lengths of the line segments.

4. The roadside object detection apparatus as claimed in claim 1 , wherein the processor is further programmed to:
detect a preceding vehicle with at least a lateral position of the preceding vehicle,
set the search range for the roadside object by translating the virtual line in a direction such that the translated virtual line is positioned between the virtual line and the preceding vehicle, and by translating the virtual line by a predetermined distance in an opposite direction, and
detect edges in the search range, to generate one or more line segments from the edges, and identify the line segment generated from the characteristic points of the roadside object in the far-off region based on at least one of the number of the line segments, and lengths of the line segments.

5. The roadside object detection apparatus as claimed in claim 1 , wherein the processor is further programmed to execute the template matching starting from a region closer to the vehicle in the search range, and when detecting the roadside object in the far-off region with the template image for the neighboring region, newly set a region including the detected roadside object as a template image to execute the template matching for a next close region to the vehicle in the search range.

6. The roadside object detection apparatus as claimed in claim 2 , wherein the processor is further programmed to estimate an existence of a sidewalk step if the number of the line segments generated from the characteristic points of the roadside object in the far-off region is two, and estimate an existence of a curbstone if the number of the line segments is three.

7. The roadside object detection apparatus as claimed in claim 3 , wherein the processor is further programmed to estimate an existence of a sidewalk step if the number of the line segments generated from the characteristic points of the roadside object in the far-off region is two, and estimate an existence of a curbstone if the number of the line segments is three.

8. The roadside object detection apparatus as claimed in claim 4 , wherein the processor is further programmed to estimate an existence of a sidewalk step if the number of the line segments generated from the characteristic points of the roadside object in the far-off region is two, and estimate an existence of a curbstone if the number of the line segments is three.

Assignees

• Takemae Yoshinao
• Watanabe Akihiro
• Toyota Motor Co Ltd

Inventors

• Takemae Yoshinao
• Watanabe Akihiro
• Ichino Fumiya

Classifications

• G06T2207/20061

  Hough transform · CPC title

• G06T2207/10021

  Stereoscopic video; Stereoscopic image sequence · CPC title

• G06T7/60

  Analysis of geometric attributes · CPC title

• B60W2554/00

  Input parameters relating to objects · CPC title

• G06T7/593

  from stereo images · CPC title