# US 2025/0384584 A1 · Online Camera Alignment Using Ensembled Features From Multiple Views

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

- Number: US 2025/0384584 A1
- Document ID: US20250384584A1
- Status: Published application
- Kind: A1
- Application number: 18/741,218
- Filed: 2024-06-12
- Published: 2025-12-18
- Earliest claimed: 2024-06-12
- Assignees: GM Global Technology Operations LLC
- Applicants: GM Global Technology Operations LLC
- Inventors: Xinyu DU; Xiaobo Guo; Mohsen Khalili; Yao Hu; Hao Yu
- Web page: https://patentgrep.com/patent/US20250384584A1

## CPC

- G06T 7/80: Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
- G06V 10/82: using neural networks
- G06V 20/56: exterior to a vehicle by using sensors mounted on the vehicle
- H04N 7/181: {for receiving images from a plurality of remote sources}
- G06T 2207/20081: Training; Learning
- G06T 2207/20084: Artificial neural networks [ANN]
- G06T 2207/30252: Vehicle exterior; Vicinity of vehicle
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION

## Abstract

A vehicle system includes one or more cameras configured to capture original images relative to a vehicle, and a control module configured to receive at least two original images from the one or more cameras of the vehicle, identify at least one target feature in the original images, select local regions of interest for the identified target feature in the original images, load view settings for each local region of interest, create one or more synthetic local images for each original image based on the local regions of interest and the loaded view settings, detect at least one feature in the one or more synthetic local images, and align the camera with an object associated with the vehicle using the detected feature. Other example vehicle systems and methods are also disclosed.

## Claims

### Claim 1

A vehicle system for a vehicle, the vehicle system comprising:
one or more cameras configured to capture original images relative to the vehicle; and
a control module in communication with the camera, the control module configured to:
receive at least two original images from the one or more cameras of the vehicle;
identify at least one target feature in the original images;
select local regions of interest for the identified target feature in the original images;
load view settings for each local region of interest, the view settings for each local region of interest including at least a size and a resolution;
create one or more synthetic local images for each original image based on the local regions of interest and the loaded view settings;
detect at least one feature in the one or more synthetic local images; and
align the camera with an object associated with the vehicle using the detected feature, wherein the object is the vehicle, the ground, or another camera of the vehicle.

### Claim 2 (depends on 1)

The vehicle system of claim 1, wherein the control module is configured to control an operation of the vehicle based on the alignment between the camera and the object.

### Claim 3 (depends on 1)

The vehicle system of claim 1, wherein the control module is configured to adjust at least one of the view settings for the local regions of interest.

### Claim 4 (depends on 3)

The vehicle system of claim 3, wherein the control module is configured to determine inlier ratios associated with a plurality of detected features and remove at least one of the view settings corresponding to the lowest inlier ratio of the inlier ratios to adjust the at least one of the view settings.

### Claim 5 (depends on 3)

The vehicle system of claim 3, further comprising a machine learning module configured to generate the view settings for each local region of interest, wherein the control module is configured to receive the view settings for each local region of interest from the machine learning module.

### Claim 6 (depends on 5)

The vehicle system of claim 5, wherein the machine learning module includes a neural network.

### Claim 7 (depends on 5)

The vehicle system of claim 5, wherein:
the control module is configured to determine inlier ratios associated with a plurality of detected features; and
the machine learning module is configured to adjust the at least one of the view settings for the local regions of interest based on the determined inlier ratios.

### Claim 8 (depends on 1)

The vehicle system of claim 1, wherein the control module is configured to generate a transformation matrix to map the detected feature between the synthetic local images and at least one of the original images.

### Claim 9 (depends on 8)

The vehicle system of claim 8, wherein the control module is configured to project the target feature back to the at least one of the original images based on the transformation matrix.

### Claim 10 (depends on 1)

A vehicle including the vehicle system of claim 1.

### Claim 11

A method for aligning one or more cameras of a vehicle with an object associated with the vehicle, the method comprising:
receiving at least two original images from the one or more cameras;
identifying at least one target feature in the original images;
selecting local regions of interest for the identified target feature in the original images;
loading view settings for each local region of interest, the view settings for each local region of interest including at least a size and a resolution;
creating one or more synthetic local images for each original image based on the local regions of interest and the loaded view settings;
detecting at least one feature in the one or more synthetic local images; and
aligning the camera with an object associated with the vehicle using the detected feature, wherein the object is the vehicle, the ground, or a second camera of the vehicle.

### Claim 12 (depends on 11)

The method of claim 11, further comprising controlling an operation of the vehicle based on the alignment between the camera and the object.

### Claim 13 (depends on 11)

The method of claim 11, further comprising adjusting at least one of the view settings for the local regions of interest.

### Claim 14 (depends on 13)

The method of claim 13, wherein adjusting the at least one of the view settings for the local regions of interest includes:
determining inlier ratios associated with a plurality of detected features; and
removing at least one of the view settings corresponding to the lowest inlier ratio of the inlier ratios to adjust the at least one of the view settings.

### Claim 15 (depends on 13)

The method of claim 13, wherein loading the view settings for each local region of interest includes generating, with a machine learning module, the view settings for each local region of interest.

### Claim 16 (depends on 15)

The method of claim 15, wherein adjusting the at least one of the view settings for the local regions of interest includes:
determining inlier ratios associated with a plurality of detected features; and
adjusting, with the machine learning module, the at least one of the view settings for the local regions of interest based on the determined inlier ratios.

### Claim 17 (depends on 11)

The method of claim 11, further comprising generating a transformation matrix to map the detected feature between the synthetic local images and at least one of the original images.

### Claim 18 (depends on 17)

The method of claim 17, further comprising projecting the target feature back to the at least one of the original images based on the transformation matrix.

### Claim 19

A non-transitory computer-readable medium storing instructions that, when executed by a control module, cause the control module to:
receive at least two original images from one or more cameras of a vehicle;
identify at least one target feature in the original images;
select local regions of interest for the identified target feature in the original images;
load view settings for each local region of interest, the view settings for each local region of interest including at least a size and a resolution;
create one or more synthetic local images for each original image based on the local regions of interest and the loaded view settings;
detect at least one feature in the one or more synthetic local images; and
align the camera with an object associated with the vehicle using the detected feature, wherein the object is the vehicle, the ground, or a second camera of the vehicle.

### Claim 20 (depends on 19)

The non-transitory computer-readable medium of claim 19, wherein the control module is further caused to control an operation of the vehicle based on the alignment between the camera and the object.