METHOD AND APPARATUS FOR CROSS-MODAL AND CROSS-LEVEL OBJECT ASSOCIATION, ELECTRONIC DEVICE AND MEDIUM

The present application relates to a method and apparatus for cross-modal and cross-level object association, an electronic device and a medium, belonging to the field of sensor data fusion. The method includes: collecting and preprocessing visual data and radar data; converting position coordinates...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Xiong, Xinli, Wan, Kailin, Xie, Binyang, Chen, Jianbin, Tan, Yu
Format: Patent
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The present application relates to a method and apparatus for cross-modal and cross-level object association, an electronic device and a medium, belonging to the field of sensor data fusion. The method includes: collecting and preprocessing visual data and radar data; converting position coordinates of radar observation objects in preprocessed radar data into corresponding pixel coordinates; acquiring first radar observation objects capable of being projected into object detection boxes in the visual data, according to the pixel coordinates corresponding to the radar observation objects; calculating a pixel distance between pixel coordinates corresponding to each of the first radar observation objects and a corresponding object detection box; associating the radar observation objects in the radar data and visual observation objects in the visual data with tracking objects; obtaining, by screening, first tracking objects that have been associated with the visual observation objects but are not associated with the radar observation objects; calculating a relative distance between each of the first radar observation objects and an own vehicle; determining a radar observation object having both a minimum relative distance and a minimum pixel distance at the same time as an associated object of a corresponding first tracking object. Collect visual data and radar data within an object scene Si Preprocess the visual data and the radar data - S2 Convert position coordinates of radar observation objects in preprocessed radar data into corresponding pixel coordinates in an , image coordinate system Acquire first radar observation objects that are capable of being projected into object detection boxes in the visual data, according to the pixel coordinates corresponding to the radar observation objects Calculate a pixel distance between pixel coordinates corresponding to each of the first radar observation objects and a corresponding object S5 detection box in the visual data Associate the radar observation objects in the preprocessed radar data and visual observation objects in preprocessed visual data with - S6 tracking objects Obtain, by screening, first tracking objects that are not associated S7 with the radar observation objects Calculate a relative distance between each of the first radar S8 observation objects and an own vehicle Determine a radar observation object having both the minimum relative distance and the minimum pixel distance at the same time as' an