# US 2019/0378281 A1 · Image-processing device, image-processing method, and recording medium

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

- Number: US 2019/0378281 A1
- Document ID: US20190378281A1
- Status: Published application
- Kind: A1
- Application number: 16/546,429
- Filed: 2019-08-21
- Published: 2019-12-12
- Earliest claimed: 2016-03-09
- Assignees: NEC Corporation
- Applicants: NEC Corporation
- Inventors: Ryo KAWAI
- Web page: https://patentgrep.com/patent/US20190378281A1

## CPC

- G06V 10/774: Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F 18/214: Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06T 5/70: Denoising; Smoothing
- G06T 7/248: {involving reference images or patches}
- G06T 7/251: {involving models}
- G06T 7/254: involving subtraction of images
- G06T 7/90: Determination of colour characteristics
- G06V 20/52: Surveillance or monitoring of activities, e.g. for recognising suspicious objects (recognising microscopic objects G06V20/69)
- G06V 40/103: {Static body considered as a whole, e.g. static pedestrian or occupant recognition}
- G06T 2207/20081: Training; Learning
- G06T 2207/20201: Motion blur correction
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

In order to produce a discriminator that has higher discrimination ability, this image-processing device is provided with a synthesis unit for synthesizing a background image and an object image the hue and/or brightness of which at least partially resembles at least a portion of the background image, a generation unit for generating a difference image between the synthesized image and the background image, and a machine learning unit for performing machine learning using the generated difference image as learning data.

## Claims

### Claim 1

An image-processing system comprising:
at least one memory; and
at least one processor configured to execute the instructions to perform:
generating a synthesized image by synthesizing a background image and an object image;
adding noise to the synthesized image;
generating a difference image between the background image and the synthesized image to which the noise has been added; and
performing machine learning using the difference image as learning data.

### Claim 2 (depends on 1)

The image-processing system according to claim 1, wherein
the noise is at least one of a pseudo-shadow, impulse noise, and Gaussian noise.

### Claim 3 (depends on 1)

The image-processing system according to claim 1, wherein
the at least one processor is configured to perform:
transforming the object image to generate a transformed object image; and
generating the synthesized image by synthesizing the background image and the transformed object image.

### Claim 4 (depends on 3)

The image-processing system according to claim 3, wherein
the at least one processor is configured to perform:
inferring a shadow part segment of an object in the synthesized image by using the synthesized image and the transformed object image;
altering a luminance of the shadow part segment of the synthesized image in relation to the synthesized image; and
adding the pseudo-shadow as noise to the synthesized image by transforming the object image.

### Claim 5

An image-processing method comprising:
generating a synthesized image by synthesizing a background image and an object image;
adding noise to the synthesized image;
generating a difference image between the background image and the synthesized image to which the noise has been added; and
performing machine learning using the difference image as learning data.

### Claim 6 (depends on 5)

The image-processing method according to claim 5, wherein
the noise is at least one of a pseudo-shadow, impulse noise, and Gaussian noise.

### Claim 7 (depends on 5)

The image-processing method according to claim 5, comprising:
transforming the object image to generate a transformed object image; and
generating the synthesized image by synthesizing the background image and the transformed object image.

### Claim 8 (depends on 7)

The image-processing method according to claim 7, comprising:
inferring a shadow part segment of an object in the synthesized image by using the synthesized image and the transformed object image;
altering a luminance of the shadow part segment of the synthesized image in relation to the synthesized image; and
adding the pseudo-shadow as noise to the synthesized image by transforming the object image.

### Claim 9

A non-transitory computer-readable recording medium storing a program to cause a computer to perform:
generating a synthesized image by synthesizing a background image and an object image;
adding noise to the synthesized image;
generating a difference image between the background image and the synthesized image to which the noise has been added; and
performing machine learning using the difference image as learning data.

### Claim 10 (depends on 9)

The non-transitory computer-readable recording medium according to claim 9, wherein
the noise is at least one of a pseudo-shadow, impulse noise, and Gaussian noise.

### Claim 11 (depends on 9)

The non-transitory computer-readable recording medium according to claim 9, wherein
the program causes the computer to perform:
transforming the object image to generate a transformed object image; and
generating the synthesized image by synthesizing the background image and the transformed object image.

### Claim 12 (depends on 11)

The non-transitory computer-readable recording medium according to claim 11, wherein
the program causes the computer to perform:
inferring a shadow part segment of an object in the synthesized image by using the synthesized image and the transformed object image;
altering a luminance of the shadow part segment of the synthesized image in relation to the synthesized image; and
adding the pseudo-shadow as noise to the synthesized image by transforming the object image.