Method identifying vehicle parking regions from aerial images, comprising: i. retrieving a first aerial image 500’ from a database; ii. extracting features from the first image 502-512; iii. detecting the presence of a car park in the first image based on the features; iv. classifying the type of parking region based on the features; v. if no vehicle parking is detected, repeating steps i-iv with a second aerial image obtained at a different time from the first aerial image; and vi. storing the classified parking region in a database comprising parking regions on Earth. The features may comprise: rows of vehicles 502; bay markings; consistent gaps between rows of vehicles or bay markings 504; roads adjacent to the row(s) of vehicles or bays 512; surrounding area characteristics 510; ground surface 506; and/or vehicle size and shape 508. Aerial image may be satellite images. Feature detection may comprise deep learning. The type of parking may be: off-street parking; on-street parking; a parking on a building; light vehicle parking; and/or a heavy vehicle parking. The different time may be selected may be a different: day; time in a day; and day with different weather conditions.
Identifying parking regions in aerial image
03.02.2021
Patent
Elektronische Ressource
Englisch
Parking lot information system using image technology for identifying available parking spaces
Europäisches Patentamt | 2015
|Unmanned aerial vehicle parking system and unmanned aerial vehicle parking method
Europäisches Patentamt | 2021
Parking apron, parking system and unmanned aerial vehicle parking method
Europäisches Patentamt | 2021
|Parking apron, parking system and unmanned aerial vehicle parking method
Europäisches Patentamt | 2023
|