Part target identification method fusing prior structure knowledge

The invention belongs to the technical field of computer vision, and particularly relates to a priori structure knowledge fused component target recognition method, which comprises the following steps of: acquiring inspection images of each area on a complex instrument in a workshop, determining a p...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: WANG CHAOZE, PENG GANG, PENG JIAXI
Format: Patent
Sprache:chi ; eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The invention belongs to the technical field of computer vision, and particularly relates to a priori structure knowledge fused component target recognition method, which comprises the following steps of: acquiring inspection images of each area on a complex instrument in a workshop, determining a pre-acquired standard image by utilizing position constraint mapping based on the acquisition position of each inspection image, and acquiring a pre-acquired standard image; acquiring a multi-channel prior semantic image of a corresponding area obtained by the standard image; a pre-constructed multi-channel semantic image input branch is adopted, the last two layers of high semantic feature images generated by the trained inspection image feature extraction network for each inspection image and a multi-channel prior semantic image corresponding to the inspection image are fused, and inspection image features are enhanced by prior structure knowledge; wherein the multi-channel semantic image input branch is construct