# US 10,504,228 B2 · Image processing system and method

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

- Number: US 10,504,228 B2
- Document ID: US10504228B2
- Status: Granted
- Kind: B2
- Application number: 15/572,587
- Filed: 2016-04-29
- Published: 2018-05-17
- Granted: 2019-12-10
- Earliest claimed: 2015-05-15
- Assignees: Sony Corporation
- Applicants: Sony Corporation
- Inventors: Alexander Gatto; Piergiorgio Sartor; Matthias Wilde; Ralf Mueller
- Web page: https://patentgrep.com/patent/US10504228B2

## CPC

- G06V 40/40: Spoof detection, e.g. liveness detection
- G06T 7/0014: {using an image reference approach}
- G06T 7/90: Determination of colour characteristics
- G06V 40/10: Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06T 7/11: Region-based segmentation
- G06T 2207/10024: Color image
- G06T 2207/30088: Skin; Dermal
- G06V 10/58: relating to hyperspectral data
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

In an image processing system a processor is configured to perform an image processing method. The method performs receiving a spectral image of a person's skin and identifying the person based on the received spectral image of the person's skin and skin reflectance information.

## Claims

### Claim 1

Image processing system, comprising:
a processor configured to
receive a spectral image of a person's skin,
generate a first skin map based on the spectral image, the first skin map identifying for each preset area of the spectral image split into preset areas whether the preset area is related to skin or not,
receive an RGB image,
generate a second skin map based on the RGB image, the second skin map identifying for each preset area of the RGB image split into preset areas whether the preset area is related to skin or not,
generate a real skin map by combining the first skin map and the second skin map, and
identify the person based on the first skin map, skin reflectance information, and the real skin map.

### Claim 2 (depends on 1)

Image processing system of claim 1, wherein
generating the real skin map includes for each preset area of the image, analyzing the preset area with respect to a skin characteristic dependent on an image type.

### Claim 3 (depends on 1)

Image processing system of claim 1, wherein the processor is further configured to
generate the first skin map based on the spectral image based on a machine learning model.

### Claim 4 (depends on 3)

Image processing system of claim 3, wherein the machine learning model is trained in advance on the basis of a set of training samples, wherein each training sample comprises a spectral image obtained from a spectral camera and a corresponding classifier which indicates if the spectral image is associated to skin or not.

### Claim 5 (depends on 1)

Image processing system of claim 1, wherein the processor is further configured to
post-filter the real skin map.

### Claim 6 (depends on 1)

Image processing system of claim 1, wherein the processor is further configured to
receive a further spectral image; and
filter the further spectral image based on the real skin map, wherein identifying the person is further based on the filtered spectral image.

### Claim 7 (depends on 1)

Image processing system of claim 1, wherein the skin reflectance information comprises skin reflectance information of each of registered users stored in a database.

### Claim 8 (depends on 1)

Image processing system of claim 1, wherein the processor is further configured to
compare the received spectral image of the person's skin with skin reflectance information of each registered user; and
identify the person as one of the registered users, if the received spectral image matches with the skin reflectance information of one of the registered users.

### Claim 9 (depends on 8)

Image processing system of claim 8, wherein the processor is further configured to
determine that the person is not a registered user, if the received spectral image of the person's skin does not match with the skin reflectance information of one of the registered users.

### Claim 10 (depends on 8)

Image processing system claim 8, wherein the processor is further configured to
register the person as registered user, if the received spectral image of the person's skin does not match with the skin reflectance information of one of the registered users.

### Claim 11 (depends on 1)

Image processing system of claim 1, wherein the spectral image comprises spectral reflectance information which is based on radiation of a predetermined wavelength or wavelength range within the visible spectrum.

### Claim 12 (depends on 1)

Image processing system of claim 1, the processor is further configured to pre-process the spectral image.

### Claim 13 (depends on 1)

Image processing system of claim 1, wherein the processor is further configured to
determine a most probable user from the registered users if the person is identified as more than one user; and
identify the person as the most probable user.

### Claim 14 (depends on 1)

Image processing system of claim 1, further comprising a spectral camera configured to provide the spectral image of the person's skin.

### Claim 15 (depends on 1)

Image processing system of claim 1, further comprising a RGB camera configured to provide an RGB image.

### Claim 16

Image processing method, comprising:
receiving a spectral image of a person's skin;
generating a first skin map based on the spectral image, the first skin map identifying for each preset area of the spectral image split into preset areas whether the preset area is related to skin or not;
receiving an RGB image;
generating a second skin map based on the RGB image, the second skin map identifying for each preset area of the RGB image split into preset areas whether the preset area is related to skin or not;
generating a real skin map by combining the first skin map and the second skin map; and
identifying the person based on the first skin map, skin reflectance information, and the real skin map.

### Claim 17

A non-transitory computer-readable storage medium storing computer-readable instructions thereon which, when executed by a computer, cause the computer to perform a method, the method comprising:
receive a spectral image of a person's skin;
generating a first skin map based on the spectral image, the first skin map identifying for each preset area of the spectral image split into preset areas whether the preset area is related to skin or not;
receiving an RGB image;
generating a second skin map based on the RGB image, the second skin map identifying for each preset area of the RGB image split into preset areas whether the preset area is related to skin or not;
generating a real skin map by combining the first skin map and the second skin map; and
identifying the person based on the first skin map, skin reflectance information, and the real skin map.