Deep learning road extraction result optimization method based on topological connectivity

The invention discloses a deep learning road extraction result optimization method based on topological connectivity. The method comprises the following steps: inputting deep learning road extraction result image data; orderly arranging edge extraction line segments of the deep learning road extract...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: SHEN ZIMO, TANIGOSHI, DAYLASER
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 discloses a deep learning road extraction result optimization method based on topological connectivity. The method comprises the following steps: inputting deep learning road extraction result image data; orderly arranging edge extraction line segments of the deep learning road extraction result; finding the position of the fracture section by using line segment phase calculation and line segment phase constraint; detecting three attributes of a road direction, a road width and a breakpoint position to which the fracture section belongs; and determining a matching relation of different fracture sections according to fracture section attributes, and connecting and optimizing the fracture sections. According to the three steps of fracture section searching, fracture section attribute determination and fracture section matching connection, the three problems that fracture sections are irregular, fracture section attribute information is difficult to extract, and fracture section matching is ambiguo