Landmark recognition device and landmark recognition method using a database storing landmark extraction information

• US 10,509,971 B2
• Filed: 03/28/2016
• Issued: 12/17/2019
• Est. Priority Date: 08/28/2015
• Status: Active Grant

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Kaneko, Alex Masuo, Yamamoto, Kenjiro
• Primary Examiner(s): Osifade, Idowu O
• Application Number:
  US15/753,767
• Publication Number:
  US 20180247139A1
• Time in Patent Office: 1,359 Days
• Field of Search: None
• CPC Class Codes: G05D 1/0246 using a video camera in combination with image processing means
• CPC Class Codes: G06T 1/0007 Image acquisition
• CPC Class Codes: G06T 2207/30256 Lane; Road marking
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06V 20/582 of traffic signs
• 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: G08G 1/0962 having an indicator mounted inside the vehicle, e.g. giving voice messages
• CPC Class Codes: G08G 1/166 for active traffic, e.g. moving vehicles, pedestrians, bikes
• CPC Class Codes: G08G 1/167 Driving aids for lane monitoring, lane changing, e.g. blind spot detection

Abstract

Landmark recognition can handle various landmarks with a low processing load and high reliability. Provided are a database, a segmentation processor, a pattern extraction processor, and a database comparison processor. The database stores landmark extraction information associated with landmark determination condition information, the landmark extraction information identifying a region including a landmark of an image, and allowing right-opposite positioning of the landmark, the landmark determination condition information used for determining landmark pattern information as the landmark. The segmentation processor sets a plurality of segmentation lines with respect to the region including the landmark of the image in a direction parallel to the landmark from a direction right opposite to the landmark, and determines a distance between intersections of the segmentation lines and a boundary of the landmark. The pattern extraction processor obtains the landmark pattern information that represents a characteristic of the pattern of the landmark as numerical value information in response to the distance between the intersections determined for each of the plurality of segmentation lines. The database comparison processor compares the landmark pattern information extracted in the pattern extraction processor with the landmark determination condition information registered in the database to identify the kind of the landmark.

Claims

1. A landmark recognition device configured to recognize plural kinds of road marks drawn on a road and/or plural kinds of road signs installed on a roadside as landmarks, comprising:

a camera mounted on a moving body for capturing an image surrounding the moving body; and

a processor configured to process the captured image to recognize a landmark, wherein

the processor includes

a database storing landmark extraction information associated with landmark determination condition information, the landmark extraction information identifying a region including a landmark of the captured image for each kind of the landmarks, and allowing right-opposite positioning of the landmark, the landmark determination condition information determining landmark pattern information as the landmark,

a segmentation processor configured to set a plurality of segmentation lines with respect to the region including the landmark of the captured image in a direction parallel to the landmark from a direction right opposite to the landmark, and determine a distance between intersections of the plurality of segmentation lines and a boundary of the landmark, where the plurality of segmentation lines are parallel with respect to one another,

a pattern extraction processor configured to obtain the landmark pattern information that represents a characteristic of the pattern of the landmark as numerical value information, in response to a distance between the intersections determined for each of the plurality of segmentation lines, and

a database comparison processor configured to compare the landmark pattern information extracted in the pattern extraction processor with the landmark determination condition information registered in the database to identify the kind of the landmark.

2. The landmark recognition device according to claim 1, wherein

the database includes

information of the region including the landmark of the captured image and information of an angle for right-opposite positioning of the landmark as the landmark extraction information for each kind of landmark,

a function for obtaining the landmark pattern information representing the characteristic of the pattern of the landmark as numerical value information for each kind of the landmark, and

a determination condition equation for evaluating a numerical value calculated by the function as the landmark determination condition information for each kind of the landmark.

3. The landmark recognition device according to claim 2, wherein

the segmentation processor is configured to

estimate a posture of the moving body relative to the landmark,

determine a region and an angle with which the plurality of segmentation lines are set on the image in accordance with the posture and the region and angle of each landmark registered in the database, and

determine a distance between the intersections of the segmentation line and the boundary of the landmark.

4. The landmark recognition device according to claim 2, wherein

the pattern extraction processor is configured to

assign coordinates of each intersection to the function of each landmark registered in the database, and

convert the coordinates to landmark pattern information.

5. A landmark recognition method for recognizing a landmark by capturing a surrounding image, wherein

landmark extraction information which is information of a region including the landmark of the captured image,

a function for obtaining landmark pattern information representing a characteristic of a pattern of the landmark as numerical value information, and

a determination condition equation for evaluating a numerical value calculated by the function as landmark determination condition information,

are provided as a set for each kind of landmark,

the landmark recognition method for recognizing the landmark, comprising:

detecting a region including the landmark with respect to the captured image;

determining a distance between intersections of a plurality of segmentation lines and a boundary of the landmark, where the plurality of segmentation lines are parallel with respect to one another;

identifying the set including the landmark extraction information of the region using a distance between the intersections determined for each of the plurality of segmentation lines;

executing the function for the identified set; and

comparing the set to the landmark determination condition information of the identified set.