# US 2019/0318490 A1 · Distance estimation of vehicle headlights

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

- Number: US 2019/0318490 A1
- Document ID: US20190318490A1
- Status: Published application
- Kind: A1
- Application number: 16/366,084
- Filed: 2019-03-27
- Published: 2019-10-17
- Earliest claimed: 2018-04-12
- Assignees: Aptiv Technologies AG
- Applicants: Aptiv Technologies Limited
- Inventors: Edin Andelic
- Web page: https://patentgrep.com/patent/US20190318490A1

## CPC

- G06V 10/764: using classification, e.g. of video objects
- G06F 18/211: Selection of the most significant subset of features
- G06F 18/2148: {characterised by the process organisation or structure, e.g. boosting cascade}
- G06F 18/24: Classification techniques
- G06T 7/11: Region-based segmentation
- G06T 7/50: Depth or shape recovery
- G06T 7/73: using feature-based methods
- G06V 10/82: using neural networks
- G06V 20/584: {of vehicle lights or traffic lights}
- B60Q 1/14: having dimming means
- G06T 2207/10016: Video; Image sequence
- G06T 2207/20081: Training; Learning
- G06T 2207/20084: Artificial neural networks [ANN]
- G06T 2207/30252: Vehicle exterior; Vicinity of vehicle
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- B60Q: ARRANGEMENT OF SIGNALLING OR LIGHTING DEVICES, THE MOUNTING OR SUPPORTING THEREOF OR CIRCUITS THEREFOR, FOR VEHICLES IN GENERAL

## Abstract

A method for determining the distance to a vehicle for use in an AHC-System includes capturing a raw image by using a camera of an AHC-System at nighttime; determining by the AHC-System that the raw image includes a headlight of a vehicle; extracting by the AHC-System an image segment of the raw image including the headlight of a vehicle; refining the image segment by applying a classifier, such as to generate a refined image including the headlight of a vehicle; building a feature vector based on the refined image; and estimating the distance to the vehicle based on the feature vector. The present invention also relates to an image processing system adapted to perform the above method.

## Claims

### Claim 1

A method for determining the distance to a vehicle for use in an AHC-System, comprising:
capturing a raw image by using a camera of an AHC-System at nighttime;
determining by the AHC-System that the raw image includes a headlight of a vehicle; and
extracting by the AHC-System an image segment of the raw image including the headlight of a vehicle;
refining the image segment by applying a classifier, such as to generate a refined image including the headlight of a vehicle;
building a feature vector based on the refined image; and
estimating the distance to the vehicle based on the feature vector.

### Claim 2 (depends on 1)

The method of claim 1, further comprising training the classifier based on a machine learning algorithm, preferably to adapt the classifier to further isolate the headlight of a vehicle.

### Claim 3 (depends on 2)

The method of claim 2, wherein the classifier is a neural network, preferably a deep-learning network, a multi-layer perceptron network, a radial basis function network or a general regression neural network.

### Claim 4 (depends on 1)

The method of any of claim 1, further comprising determining a distance estimator by using training data, wherein the distance estimator is adapted to estimate the distance to the vehicle based on the feature vector.

### Claim 5 (depends on 4)

The method of claim 4, wherein the training data includes ground truth data defining the relation between different feature vectors and different distances to a vehicle, wherein the distances to the vehicle are in the range 10 m to 400 m.

### Claim 6 (depends on 4)

The method of claim 4, wherein the training data includes ground truth data defining the relation between different feature vectors and different distances to a vehicle, wherein the distances to the vehicle are in the range 50 m to 200 m.

### Claim 7 (depends on 4)

The method of claim 4, wherein the training data includes ground truth data defining the relation between different feature vectors and different distances to a vehicle, wherein the distances to the vehicle are in the range 100 m to 150 m.

### Claim 8 (depends on 4)

The method of claim 4, wherein the distance estimator is based on a regression model.

### Claim 9 (depends on 4)

The method of claim 4, wherein the distance estimator is based on a machine learning algorithm, preferably a neural network, further preferred wherein the distance estimator provides a discrete or continuous value corresponding to the estimated distance to the vehicle.

### Claim 10 (depends on 9)

The method of claim 9, wherein the classifier corresponds to a first neural network, and the distance estimator corresponds to a second neural network, and wherein the number of neurons in the first neural network is at least 2 times larger than the number of neurons in the second neural network.

### Claim 11 (depends on 9)

The method of claim 9, wherein the classifier corresponds to a first neural network, and the distance estimator corresponds to a second neural network, and wherein the number of neurons in the first neural network is at least 10 times larger than the number of neurons in the second neural network.

### Claim 12 (depends on 1)

The method of any of claim 1, wherein the refined image corresponds to an image excerpt of the image segment.

### Claim 13 (depends on 1)

The method of any of claim 1, further comprising triggering the step of refining the image segment by applying a classifier if the AHC-System determines that the image segment includes a headlight of a vehicle.

### Claim 14 (depends on 1)

The method of any of claim 1, wherein building a feature vector based on the refined image comprises determining: the number of pixels or the pixel area defined by the refined image, the aver-age pixel intensity in the refined image, or the standard deviation of pixel intensities in the refined image.

### Claim 15 (depends on 1)

The method of any of claim 1, wherein building a feature vector based on the refined image comprises determining: the position of the refined image relative to the image segment or relative to the raw image.

### Claim 16 (depends on 1)

The method of any of claim 1, wherein the feature vector does not contain the refined image or an image excerpt of the refined image.

### Claim 17 (depends on 1)

The method of any of claim 1, wherein the raw image is captured by the AHC-System by using a monocular camera, preferably a video camera, further preferred a front camera of a vehicle.

### Claim 18

An image processing system, comprising:
an AHC-System adapted to:
capture a raw image by using a camera at nighttime;
determine that the raw image includes a headlight of a vehicle; and
extract an image segment of the raw image including the headlight of a vehicle;
characterized by:
an image processing means adapted to:
refine the image segment by applying a classifier, such as to generate a refined image including the headlight of a vehicle;
build a feature vector based on the refined image; and
estimate the distance to the vehicle based on the feature vector.