Object recognition apparatus and noise removal method

Patent metadata

Field	Value
Publication number	US-10502832-B2
Application number	US-201615055239-A
Country	US
Kind code	B2
Filing date	Feb 26, 2016
Priority date	Feb 27, 2015
Publication date	Dec 10, 2019
Grant date	Dec 10, 2019

Abstract

In an object recognition apparatus, a range point acquirer irradiates each of irradiation areas arranged in a horizontally and vertically extending grid and forming a detection area for detecting a target with laser light and receives the reflected light from the respective irradiation areas, thereby acquiring a plurality of range points representing per-irradiation area coordinates of the target. A noise remover is configured to, based on either or both of degrees of angle proximity and degrees of distance proximity between a plurality of subject range points to be determined whether to be a noise point, of the plurality of range points, as viewed from a reference point, determine whether or not each of the subject range points is a noise point, and remove the noise point from the plurality of range points. An object recognizer uses the plurality of range points other than the noise points to recognize the object.

First claim

What is claimed is:

1. An object recognition apparatus for recognizing an object as a target, the apparatus comprising:
a range point acquirer configured to irradiate each of irradiation areas arranged in a horizontally and vertically extending grid and forming a detection area for detecting a target with laser light and receive the reflected light from the respective irradiation areas, thereby acquiring a plurality of range points representing per-irradiation area coordinates of the target;
a noise remover configured to, based on either or both of degrees of angle proximity and degrees of distance proximity between a plurality of subject range points to be determined whether to be a noise point, of the plurality of range points, as viewed from a reference point, determine whether or not each of the subject range points is a noise point, and remove the noise points from the plurality of range points; and
an object recognizer configured to use the plurality of range points other than the noise points to recognize the object,
wherein the noise remover is configured to calculate a co-linearity indicative of a degree of proximity of a first one of the plurality of subject range points to a line passing through a second one of the subject range points and the reference point, and if the co-linearity is equal to or greater than a predetermined value, then determine whether or not each of the subject range points is likely to be a noise point.

2. The apparatus of claim 1 , wherein the noise remover is configured to calculate angles between lines passing through the reference point and the respective subject range points, and determine that the co-linearity becomes higher as the angle decreases.

3. The apparatus of claim 1 , wherein the noise remover is configured to, based on a continuity in distance between range points in adjacent irradiation areas, of the plurality of subject range points, determine whether or not each of the subject range points is likely to be a noise point.

4. The apparatus of claim 1 , wherein the noise remover is configured to, based on a horizontal continuity in distance, determine whether or not each of the subject range points is likely to be a noise point.

5. The apparatus of claim 4 , wherein the noise remover is configured to, based on a continuity in distance between range points in horizontally adjacent irradiation areas, of the plurality of subject range points, determine whether or not each of the subject range points is likely to be a noise point.

6. The apparatus of claim 1 , wherein the noise remover is configured to, based on a vertical continuity in distance, determine whether or not each of the subject range points is likely to be a noise point.

7. The apparatus of claim 6 , wherein the noise remover is configured to, based on a continuity in distance between range points in vertically adjacent irradiation areas, of the plurality of subject range points, determine whether or not each of the subject range points is likely to be a noise point.

8. The apparatus of claim 1 , wherein the noise remover is configured to determine whether or not there is a range point acquired in the past within a predetermined reference distance from a subject range point to be determined whether to be a noise point, and if there is no range point acquired in the past within the reference distance from the subject range point, then determine that the subject range point is likely to be a noise point.

9. The apparatus of claim 1 , wherein the noise remover is configured to calculate an evaluation value indicative of a likelihood that a range point is a noise point, and if the evaluation value of the range point is equal to or greater than a threshold, then remove the range point, as a noise point.

10. An object recognition apparatus for recognizing an object as a target, the apparatus comprising:
a range point acquirer configured to irradiate each of irradiation areas arranged in a horizontally and vertically extending grid and forming a detection area for detecting a target with laser light and receive the reflected light from the respective irradiation areas, thereby acquiring a plurality of range points representing per-irradiation area coordinates of the target;
a noise remover configured to, based on either or both of degrees of angle proximity and degrees of distance proximity between a plurality of subject range points to be determined whether to be a noise point, of the plurality of range points, as viewed from a reference point, determine whether or not each of the subject range points is a noise point, and remove the noise points from the plurality of range points; and
an object recognizer configured to use the plurality of range points other than the noise points to recognize the object,
wherein the noise remover is configured to determine whether or not there is a range point acquired in the past within a predetermined reference distance from a subject range point to be determined whether to be a noise point, and if there is no range point acquired in the past within the reference distance from the subject range point, then determine that the subject range point is likely to be a noise point, and
the noise remover is configured to determine whether or not there is a cluster of range points acquired in the past within a reference distance from a subject cluster of range points to be determined whether to be a noise point, and if there is no cluster of range points acquired in the past within the reference distance from the subject cluster, then determine the subject cluster of range points is likely to be a noise point.

11. An object recognition apparatus for recognizing an object as a target, the apparatus comprising:
a range point acquirer configured to irradiate each of irradiation areas arranged in a horizontally and vertically extending grid and forming a detection area for detecting a target with laser light and receive the reflected light from the respective irradiation areas, thereby acquiring a plurality of range points representing per-irradiation area coordinates of the target;
a noise remover configured to, based on either or both of degrees of angle proximity and degrees of distance proximity between a plurality of subject range points to be determined whether to be a noise point, of the plurality of range points, as viewed from a reference point, determine whether or not each of the subject range points is a noise point, and remove the noise points from the plurality of range points; and
an object recognizer configured to use the plurality of range points other than the noise points to recognize the object,
wherein the noise remover is configured to calculate an evaluation value indicative of a likelihood that a range point is a noise point, and if the evaluation value of the range point is equal to or greater than a threshold, then remove the range point, as a noise point, and
the noise remover is configured to calculate single or plural evaluation values, and remove a subject range point to be determined whether to be a noise point, for which a weighted sum of the single or plural evaluation values is equal to or greater than a threshold, as a noise point.

12. A noise removal method for removing noise points from a plurality of range points, the method comprising steps of:
irradiating each of irradiation areas arranged in a horizontally and vertically extending grid and forming a detection area for detecting a target with laser light and receiving the reflected light from the respective irradiation areas, thereby acquiring a plurality of range points representing per-irradiation area coordinates of the target;
based on either or both of degre

Assignees

• Denso Corp
• Toyota School Found

Inventors

• Tehrani Niknejad Hossein
• Sakai Utsushi
• Ogawa Takashi
• Mita Seiichi
• Xie Qiwei

Classifications

• G01S7/4802

  using analysis of echo signal for target characterisation; Target signature; Target cross-section · CPC title

• G01S17/931

  of land vehicles · CPC title

• G01S17/42 Primary

  Simultaneous measurement of distance and other co-ordinates (indirect measurement G01S17/46) · CPC title

• G01S7/4808

  Evaluating distance, position or velocity data · CPC title

• G01S17/936

  Physics · mapped topic