Accident analysis shows that intersections are a focal point for accidents in urban areas. Due to that, traffic monitoring at intersections has attracted much attention. A major part of the Ko-PER project, which is part of the research initiative Ko-FAS is the infrastructure based perception of all dynamic objects in an intersection. To receive a consistent 3D bird's eye view of the intersection the data of multiple spatial distributed laser scanners have to be transformed into one common coordinate system. Due to the overlapping field of views and the highmounting position of the sensor network 3D shape information of the objects in the intersection can be obtained. The 3D profiles open up new possibilities in object recognition with laser scanners. One approach to use the shape information is to model it with Active Shape Models (ASM). Based on the stability of the sensor positions and the background of the scene the measurements resulting of dynamic objects can be extracted using background subtraction. After the segmentation of the remaining data points object hypotheses are generated, which can be verified with a ASM. In addition to the training of the ASM the publication deals with robust point set registration methods, which are essential for the use of an ASM for classification.
High-precision 3D traffic recognition at intersections
Hochgenaue, dreidimensionale Verkehrsbeurteilung an Kreuzung-/Einmündungsbereichen
2010
6 Seiten, 13 Bilder, 1 Tabelle, 9 Quellen
(nicht paginiert)
Aufsatz (Konferenz)
Datenträger
Englisch
Intersections without traffic signals
SLUB | 1988
|Handling traffic at intersections
Engineering Index Backfile | 1934
|Eliminating street traffic intersections
Engineering Index Backfile | 1930
|