Specific area detection device

Patent metadata

Field	Value
Publication number	US-11301701-B2
Application number	US-202016939347-A
Country	US
Kind code	B2
Filing date	Jul 27, 2020
Priority date	Aug 7, 2019
Publication date	Apr 12, 2022
Grant date	Apr 12, 2022

Abstract

A specific area detection device that detects a specific area in an imaging area based on a captured image includes: an estimation unit configured to estimate a plurality of points and a direction of a straight line connecting two predetermined points among the plurality of points from a captured image using a learning model created by learning about a specific area defined by a predetermined number of the points in an imaging area using a captured image for learning; and a detection unit configured to detect the specific area by classifying the plurality of points for each specific area based on the direction.

First claim

What is claimed is:

1. A specific area detection device that detects a specific area in an imaging area based on a captured image, the specific area detection device comprising:
a processor configured to implement
an estimation unit configured to estimate a plurality of points and directions of a straight line connecting two predetermined points among the plurality of points from the captured image using a learning model created by learning about a specific area defined by a predetermined number of the plurality of points in an imaging area using a previously captured image for learning; and
a detection unit configured to detect the specific area by classifying the plurality of points for each specific area based on one of the directions of the straight line connecting the two predetermined points,
wherein the specific area is a parking area and the directions of the straight line connecting the two predetermined points is different than a line demarcating the parking area in the captured image.

2. The specific area detection device according to claim 1 , wherein
the estimation unit estimates the plurality of points and the directions of the straight line connecting the two predetermined points from a captured image using a second learning model created by learning about a quadrangular parking section defined by four corner points as the specific area, and
the detection unit detects the quadrangular parking section by classifying the plurality of points for the quadrangular parking section based on one of the directions.

3. The specific area detection device according to claim 2 , further comprising:
a display control unit configured to cause the parking section detected by the detection unit to be displayed on a display unit.

4. A specific area detection device that detects a specific area in an imaging area based on a captured image, the specific area detection device comprising:
a processor configured to implement:
an estimation unit configured to estimate a plurality of points and a directions of a straight line connecting two predetermined points among the plurality of points from a captured image using a learning model created by learning about a specific area defined by a predetermined number of the points in an imaging area using a previously captured image for learning; and
a detection unit configured to detect the specific area by classifying the plurality of points for each specific area based on one of the directions of the straight line connecting the two predetermined points, wherein
the estimation unit estimates the plurality of points and the directions of the straight line connecting the two predetermined points from a captured image using a second learning model created by learning about a quadrangular parking section defined by four corner points as the specific area,
the detection unit detects the quadrangular parking section by classifying the plurality of points for the parking section based on one of the directions,
the estimation unit is configured to:
when estimating the plurality of points, estimate the plurality of points with each of the plurality of points classified into any of four kinds of points including a first point, a second point, a third point, and a fourth point in a predetermined rotation direction based on a direction of vehicle approach toward the parking section, and when estimating the directions of the straight line connecting the two predetermined points, estimate a directions of a straight line connecting the first point and the second point which are estimated to belong to a same a quadrangular parking section, and
the detection unit is configured to
classify the plurality of points for each parking section based on the directions by searching for a combination of the first point and the second point with a minimum cost using a cost matrix corresponding to a bipartite graph of a plurality of the first points and a plurality of the second points, and
when a predetermined first point and a predetermined second point are selected, calculate a cost based on three points including the predetermined first point, the predetermined second point, and an intersection point of a perpendicular line drawn from the predetermined second point with respect to the straight line connecting the first point and the second point indicating the directions estimated in relation to the predetermined first point and the straight line.

5. The specific area detection device according to claim 4 , wherein
the detection unit searches for the combination of the first point and the second point with the minimum cost by applying a Hungarian method to the cost matrix.

Assignees

• Aisin Corp

Inventors

• Kokubo Yoshihito
• Suetsugu Yoshihisa
• Adachi Jun
• Watanabe Hiroyuki
• Ueda Koki

Classifications

• G06V10/764 Primary

  using classification, e.g. of video objects · CPC title

• G06V10/82

  using neural networks · CPC title

• G06V20/586

  of parking space · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G08G1/0125

  Traffic data processing · CPC title