DE102019000439A1 - Method for calibrating a vehicle camera - Google Patents

Abstract

The invention relates to a method for calibrating a vehicle camera (2) of a vehicle, wherein images (B) of a vehicle environment are detected by means of the vehicle camera (2). According to the invention, the images (B) are detected during a standstill of the vehicle and at least one distinctive environment object is selected as the image motif, wherein image features are extracted from the images (B), from the extracted image features an environment pattern (U) specific to a respective vehicle environment is formed , At least one formed specific environmental pattern (U) with at least one stored in a database (4) target environmental pattern (Us) is compared and determined by a determined during the comparison between the formed specific environmental pattern (U) and the target environmental pattern (Us ) the vehicle camera (2) is calibrated.

Description

The invention relates to a method for calibrating a vehicle camera according to the preamble of claim 1.

• - Bereitstellen eines Fahrzeugs mit einer Kamera;
• - Mehrmaliges Ermitteln der extrinsischen Parameter der Kamera während einer Messfahrt, wobei während jedes Ermittelns der extrinsischen Parameter der Kamera Sensordaten mittels eines Sensors erzeugt werden;
• - Ablegen der ermittelten Parameter und der jeweils zugehörigen Sensordaten in einer Datenbank;
• - Trainieren eines Convolutional Neural Networks basierend auf den in der Datenbank abgelegten extrinsischen Parametern und basierend auf den Sensordaten; und
• - Online-Bestimmen der extrinsischen Parameter der Kamera mittels des trainierten Convolutional Neural Networks.

From the DE 10 2016 223 422 A1 For example, a method for automatically determining extrinsic parameters of a camera of a vehicle is known by the following steps:

• - Providing a vehicle with a camera;
• Repeatedly determining the extrinsic parameters of the camera during a test drive, wherein sensor data is generated by means of a sensor during each determination of the extrinsic parameters of the camera;
• - Store the determined parameters and the corresponding sensor data in a database;
• - training a convolutional neural network based on the extrinsic parameters stored in the database and based on the sensor data; and
• - Online determination of the extrinsic parameters of the camera using the trained Convolutional Neural Networks.

In this case, the camera forms the sensor, wherein each time the extrinsic parameters of the camera are acquired an image is generated by means of the camera and a corresponding image data set, the extrinsic parameters determined and the associated image data set are stored in the database and the training of the convolutional neural network is based on the stored in the database extrinsic parameters and the image data sets.

The invention is based on the object to provide a comparison with the prior art improved method for calibrating a vehicle camera.

The object is achieved by a method having the features specified in claim 1.

In the method for calibrating a vehicle camera of a vehicle, images of a vehicle environment are acquired by means of the vehicle camera.

According to the invention, the images are captured during a standstill of the vehicle and at least one distinctive environmental object is selected as the image motif. From the images, image features are extracted, wherein from the extracted image features a specific for a particular vehicle environment environment pattern is formed. Furthermore, at least one specific environmental pattern formed is compared with at least one target environmental pattern stored in a database, and the vehicle camera is calibrated on the basis of a deviation determined between the specific environmental pattern formed and the target environmental pattern during the comparison.

By means of the method, self-calibrations of the vehicle camera can be carried out at regular intervals in order, for example, to fulfill requirements for applications in an autonomous driving operation of the vehicle. In this case, a regular determination of the deviations between formed specific environmental patterns and desired environmental patterns allows a regular calibration of the vehicle camera and thus a high level of reliability of the same as well as freedom from errors of image data captured by the vehicle camera. Thus, high reliability and error-free operation of assistance systems which use image data acquired by the camera for their operation can be achieved. In doing so, original variables can be taken into account when determining the deviations between formed specific environmental patterns and desired environmental patterns.

Advantageous embodiments of the invention are the subject of the dependent claims.

Embodiments of the invention will be explained in more detail below with reference to a drawing.

• 1 schematisch ein Blockschaltbild einer Vorrichtung zum Kalibrieren einer Fahrzeugkamera.

Showing:

• 1 schematically a block diagram of a device for calibrating a vehicle camera.

In the only one 1 is a block diagram of a possible embodiment of a device 1 for calibrating a vehicle camera 2 of a vehicle.

For calibrating the vehicle camera 2 be by means of this during a stoppage of the vehicle, for example, while waiting for the same at a road intersection and / or a traffic light, pictures B detected a vehicle environment.

In this case, at least one distinctive environment object is selected as the image motif, for example a place of interest stored in a navigation system, also referred to as point of interest (POI for short). Alternatively or additionally, the at least one distinctive environment object is detected at such a location of interest. Such places of interest include, among others, distinctive ones Buildings and intersections with fixed stops on which the pictures B always from the same position are detectable. For example, it is of great advantage that the vehicle camera 2 when the vehicle first waits at a road intersection, it has a clear view of the vehicle surroundings in the area of the intersection.

The from the vehicle camera 2 captured images B are sent to a processing unit 3 transmitted, which from the pictures B By means of edge extraction, segmentation and / or determination of pixel values, image features are extracted and from these an environment pattern specific for a respective vehicle environment U forms.

The specific environmental pattern U is subsequently by means of the processing unit 3 with at least one in a database 4 stored target environmental pattern Us, wherein based on a determined during the comparison deviation between the formed specific environmental pattern U and the target environmental pattern Us the vehicle camera 2 is calibrated. The calibration can be carried out during the standstill of the vehicle and / or during a subsequent to the stoppage of the vehicle movement of the vehicle.

LIST OF REFERENCE NUMBERS

1

device

2

vehicle camera

3

processing unit

4

Database

B

image

U

environment pattern

U _S

Target environment pattern

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• DE 102016223422 A1 [0002]

Claims (4)

A method for calibrating a vehicle camera (2) of a vehicle, wherein by means of the vehicle camera (2) images (B) of a vehicle environment are detected, characterized in that - the images (B) are detected during a stoppage of the vehicle, - as a picture motif at least one marked environment object is selected, - image features are extracted from the images (B), - an environment pattern (U) specific to a respective vehicle environment is formed from the extracted image features, - at least one formed specific environmental pattern (U) having at least one in a database ( 4) stored target environmental pattern (Us) is compared and - the vehicle camera (2) is calibrated based on a determined during the comparison deviation between the formed specific environmental pattern (U) and the target environmental pattern (Us). Method according to Claim 1 , characterized in that the calibration is performed during a subsequent to the stoppage of the vehicle movement of the vehicle. Method according to Claim 1 or 2 , characterized in that the image features are extracted from the images (B) by means of - edge extraction, - segmentation and / or - determination of pixel values. Method according to one of the preceding claims, characterized in that a location stored in a navigation system of interest and / or the at least one distinctive environment object is detected at a location of interest as a distinctive environment object.

Images