# US 2026/0311347 A1 · Object recognition method and object recognition device

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0311347 A1
- Document ID: US20260311347A1
- Status: Published application
- Kind: A1
- Application number: 19/482,288
- Filed: 2023-05-26
- Published: 2026-10-08
- Earliest claimed: 2023-05-26
- Assignees: Nissan Motor Co., Ltd.
- Applicants: Nissan Motor Co., Ltd.
- Inventors: Akinobu IKEDA
- Web page: https://patentgrep.com/patent/US20260311347A1

## CPC

- G06T 7/248: {involving reference images or patches}
- G06T 7/254: involving subtraction of images
- G06T 2207/10016: Video; Image sequence
- G06T 2207/20224: Image subtraction
- G06T 2207/30232: Surveillance
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

A moving object recognition method by a computer includes acquiring a plurality of surrounding images in a time series obtained by capturing surroundings a plurality of times in a time series. A row image is extracted at a prescribed position in a vertical direction from each of the plurality of surrounding images in the time series. A time-series image is generated in which the extracted row images are arranged in the vertical direction in a time series order. A moving object is recognized moving in the surroundings from the plurality of surrounding images in the time series based on a difference between the row images of the time-series image in a horizontal position of a feature point at which a feature quantity in the row images varies.

## Claims

### Claim 1

A moving object recognition method executed by a computer, the moving object recognition method comprising:
acquiring a plurality of surrounding images in a time series obtained by capturing surroundings a plurality of times in a time series;
extracting row images at two or more prescribed positions in a vertical direction different from each other, from each of the acquired plurality of surrounding images in the time series;
generating, for each of the two or more prescribed positions, a time-series image in which the extracted row images are arranged in the vertical direction in a time series order;
grouping the feature points of the time-series images that move between the row images and at which feature quantities in the row images vary, based on moving directions of horizontal positions of the feature points and on parameters corresponding to straight line components approximating time-series changes of the horizontal positions;
clustering two of the feature points into a same group when a moving speed difference between the horizontal positions of the two of the feature points that move between the row images is less than a prescribed value and a difference between the parameters of the two of the feature points is within a prescribed range; and
recognizing a moving object moving in the surroundings from the plurality of surrounding images in the time series based on a difference between the row images of each of the time-series images in the horizontal positions of the feature points clustered into the same group.

### Claim 2 (depends on 1)

The moving object recognition method according to claim 1, further comprising:
extracting the feature point from each of the row images of each of the time-series images; and
detecting the difference between the row images of each of the time-series images in the horizontal position of the extracted feature point.

### Claim 3 (depends on 1)

The moving object recognition method according to claim 1, further comprising
extracting an edge portion in each of the row images of each of the time-series images as the feature point.

### Claim 4 (depends on 1)

The moving object recognition method according to claim 1, further comprising
extracting the feature point in each of the row images of each of the time-series images based on a negative variation of the feature quantity in a horizontal direction.

### Claim 5 (depends on 1)

The moving object recognition method according to claim 1, further comprising
recognizing the moving object based on a moving speed of the feature point whose horizontal position moves between the row images of each of the time-series images.

### Claim 6

(canceled)

### Claim 7 (depends on 1)

The moving object recognition method according to claim 1, further comprising
recognizing the moving object when a score, which becomes higher as the grouped feature points in each of the time-series images extend longer in the vertical direction, is equal to or greater than a prescribed determination threshold.

### Claim 8 (depends on 1)

The moving object recognition method according to claim 1, further comprising
recognizing the moving object in a surrounding region of a boundary of a structure present in the surrounding image.

### Claim 9 (depends on 8)

The moving object recognition method according to claim 8, further comprising
arranging, in the time-series images, each of the row images at a position offset from the two or more prescribed positions in the vertical direction based on a moving speed of the structure on the surrounding image.

### Claim 10

A moving object recognition device comprising a computer, wherein the computer is configured to:
acquire a plurality of surrounding images in a time series obtained by capturing the surroundings a plurality of times in a time series;
extract row images at two or more prescribed positions in a vertical direction different from each other, from each of the acquired plurality of surrounding images in the time series;
generate, for each of the two or more prescribed positions, a time-series image in which the extracted row images are arranged in the vertical direction in a time-series order;
group feature points of the time-series images that move between the row images and at which the feature quantities in the row images vary, based on moving directions of horizontal positions of the feature points and on parameters corresponding to straight line components approximating time-series changes of the horizontal positions;
cluster two of the feature points into a same group when a moving speed difference between the horizontal positions of the two of the feature points that move between the row images is less than a prescribed value and a difference between the parameters of the two of the feature points is within a prescribed range; and
recognize a moving object moving in the surroundings from the plurality of surrounding images in the time series based on a difference between the row images of each of the time-series images in the horizontal positions of the feature points clustered into the same group.
