US 12,450,920 B2
Method of detecting at least one traffic lane marking and/or road marking in at least one digital image representation
Denis Tananaev, Sindelfingen (DE)
Assigned to ROBERT BOSCH GMBH, Stuttgart (DE)
Filed by Robert Bosch GmbH, Stuttgart (DE)
Filed on Jan. 12, 2023, as Appl. No. 18/153,753.
Claims priority of application No. 10 2022 200 515.5 (DE), filed on Jan. 18, 2022; and application No. 10 2022 214 341.8 (DE), filed on Dec. 22, 2022.
Prior Publication US 2023/0230395 A1, Jul. 20, 2023
Int. Cl. G06V 20/56 (2022.01); G06T 9/00 (2006.01); G06V 10/82 (2022.01)
CPC G06V 20/588 (2022.01) [G06T 9/00 (2013.01); G06V 10/82 (2022.01)] 8 Claims
OG exemplary drawing
 
1. A method of detecting at least one traffic lane and/or a roadway marking in at least one digital image representation based on sensor data obtained from at least one environmental sensor of a system of a vehicle, the method comprising:
a) obtaining a plurality of digital image representations each containing a plurality of features representing respective image content;
b) applying a bird's eye view transformation to the obtained digital image representations, wherein each of the digital image representations is transformed separately so as to create transformed digital image representations; and
c) performing a consolidation of the transformed digital image representations to obtain a consolidated digital image representation;
wherein each of the digital image representations includes a feature compilation or is provided as the feature compilation, so that an artificial neural network is used to process the feature compilation, so that the bird's eye view transformation is applied to the feature compilation,
wherein the feature compilation is a feature map, which is a description of the digital image representation which is already partially analyzed and/or in which features are detected in the digital image representation and stored in the form of the feature map,
wherein the digital image representation is a combination of raw image data and the feature compilation, and
wherein detection of traffic lanes and/or roadway markings is performed only after the bird's eye view transformation and the consolidation of steps b) and c).