# US 12,602,857 B2 · Generating image data

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

- Number: US 12,602,857 B2
- Document ID: US12602857B2
- Status: Granted
- Kind: B2
- Application number: 18/596,505
- Filed: 2024-03-05
- Published: 2025-09-11
- Granted: 2026-04-14
- Earliest claimed: 2024-03-05
- Assignees: QUALCOMM Incorporated
- Applicants: QUALCOMM Incorporated
- Inventors: Yan Deng; Michel Adib Sarkis; Chieh-Ming Kuo; Ze Zhang; Ning Bi
- Web page: https://patentgrep.com/patent/US12602857B2

## CPC

- G06T 15/04: Texture mapping
- G06T 17/00: Three-dimensional [3D] modelling for computer graphics
- G06V 10/44: Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06V 10/806: {of extracted features}
- H04N 13/117: the virtual viewpoint locations being selected by the viewers or determined by viewer tracking
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION

## Abstract

Systems and techniques are described herein for generating image data. For instance, a method for generating image data is provided. The method may include selecting enrollment images from among a plurality of enrollment images of a subject based on a target viewing angle, wherein each enrollment image of the plurality of enrollment images represents the subject as viewed from a different respective viewing angle; determining, from the selected enrollment images, appearance features representing at least one characteristic of the subject in the selected enrollment images; combining the appearance features based on a three-dimensional geometry of the subject to generate combined appearance features; and generating an image of the subject from the target viewing angle based on a texture image of the subject and the combined appearance features.

## Claims

### Claim 1

An apparatus for generating image data, the apparatus comprising:
at least one memory; and
at least one processor coupled to the at least one memory and configured to:
select enrollment images from among a plurality of enrollment images of a subject based on a target viewing angle, wherein each enrollment image of the plurality of enrollment images represents the subject as viewed from a different respective viewing angle;
process the plurality of enrollment images using machine-learning model to generate image features;
pool the image features based on a mask indicative of at least one characteristic to generate characteristic features;
transform the characteristic features based on the selected enrollment images to generate appearance features representing the at least one characteristic of the subject in the selected enrollment images;
combine the appearance features based on a three-dimensional geometry of the subject to generate combined appearance features; and
generate an image of the subject from the target viewing angle based on a texture image of the subject and the combined appearance features.

### Claim 2 (depends on 1)

The apparatus of claim 1, wherein the selected enrollment images are selected based on a relationship between respective viewing angles of the selected enrollment images and the target viewing angle.

### Claim 3 (depends on 1)

The apparatus of claim 1, wherein the appearance features are generated by an appearance-code extractor.

### Claim 4 (depends on 1)

The apparatus of claim 1, wherein, to combine the appearance features, the at least one processor is configured to process the appearance features using a machine-learning model to generate the combined appearance features.

### Claim 5 (depends on 1)

The apparatus of claim 1, wherein, to combine the appearance features, the at least one processor is configured to process the appearance features and a mask indicative of the at least one characteristic using a machine-learning model to generate the combined appearance features.

### Claim 6 (depends on 1)

The apparatus of claim 1, wherein:
the subject is a head of a person; and
the at least one characteristic is hair of the person.

### Claim 7 (depends on 1)

The apparatus of claim 1, wherein:
the texture image of the subject comprises an image of the subject without the at least one characteristic from the target viewing angle; and
the generated image of the subject includes the at least one characteristic.

### Claim 8 (depends on 1)

The apparatus of claim 1, wherein the image of the subject is generated further based on normals of planes of the at least one characteristic.

### Claim 9

A method for generating image data, the method comprising:
selecting enrollment images from among a plurality of enrollment images of a subject based on a target viewing angle, wherein each enrollment image of the plurality of enrollment images represents the subject as viewed from a different respective viewing angle;
processing the plurality of enrollment images using machine-learning model to generate image features;
pooling the image features based on a mask indicative of at least one characteristic to generate characteristic features;
transforming the characteristic features based on the selected enrollment images to generate appearance features representing the at least one characteristic of the subject in the selected enrollment images;
combining the appearance features based on a three-dimensional geometry of the subject to generate combined appearance features; and
generating an image of the subject from the target viewing angle based on a texture image of the subject and the combined appearance features.

### Claim 10 (depends on 9)

The method of claim 9, wherein the selected enrollment images are selected based on a relationship between respective viewing angles of the selected enrollment images and the target viewing angle.

### Claim 11 (depends on 9)

The method of claim 9, wherein the appearance features are generated by an appearance-code extractor.

### Claim 12 (depends on 9)

The method of claim 9, wherein combining the appearance features comprises processing the appearance features using a machine-learning model to generate the combined appearance features.

### Claim 13 (depends on 9)

The method of claim 9, wherein combining the appearance features comprises processing the appearance features and a mask indicative of the at least one characteristic using a machine-learning model to generate the combined appearance features.

### Claim 14 (depends on 9)

The method of claim 9, wherein:
the subject is a head of a person; and
the at least one characteristic is hair of the person.

### Claim 15 (depends on 9)

The method of claim 9, wherein:
the texture image of the subject comprises an image of the subject without the at least one characteristic from the target viewing angle; and
the generated image of the subject includes the at least one characteristic.

### Claim 16 (depends on 9)

The method of claim 9, wherein the image of the subject is generated further based on normals of planes of the at least one characteristic.

### Claim 17

A non-transitory computer-readable storage medium having stored thereon instructions that, when executed by at least one processor, cause the at least one processor to:
select enrollment images from among a plurality of enrollment images of a subject based on a target viewing angle, wherein each enrollment image of the plurality of enrollment images represents the subject as viewed from a different respective viewing angle;
process the plurality of enrollment images using machine-learning model to generate image features;
pool the image features based on a mask indicative of at least one characteristic to generate characteristic features;
transform the characteristic features based on the selected enrollment images to generate appearance features representing the at least one characteristic of the subject in the selected enrollment images;
combine the appearance features based on a three-dimensional geometry of the subject to generate combined appearance features; and
generate an image of the subject from the target viewing angle based on a texture image of the subject and the combined appearance features.

### Claim 18 (depends on 17)

The non-transitory computer-readable storage medium of claim 17, wherein the selected enrollment images are selected based on a relationship between respective viewing angles of the selected enrollment images and the target viewing angle.