Details
| Originalsprache | Englisch |
|---|---|
| Seiten | 51-56 |
| Seitenumfang | 6 |
| Publikationsstatus | Veröffentlicht - 2007 |
| Veranstaltung | Photogrammetric Image Analysis, PIA 2007 - Munich, Deutschland Dauer: 19 Sept. 2007 → 21 Sept. 2007 |
Konferenz
| Konferenz | Photogrammetric Image Analysis, PIA 2007 |
|---|---|
| Land/Gebiet | Deutschland |
| Ort | Munich |
| Zeitraum | 19 Sept. 2007 → 21 Sept. 2007 |
Abstract
This paper deals with road extraction of high resolution aerial images of suburban scenes based on segmentation using the Normalized Cuts algorithm. The aim of our project is the extraction of roads for the assessment of a road database, however, this paper is restricted to road extraction. The segmentation as our basic step is designed to yield a good division between road areas and the surroundings. We use the Normalized Cuts algorithm, which is a graph-based approach that divides the image on the basis of pixel similarities. The definition of these similarities can incorporate several features, which is necessary for the segmentation in complex surroundings such as built-up areas. The features used for segmentation comprise colour, hue, edges and road colour derived with prior information about the position of the centerline from the database. The initial segments have to be grouped due to an enforced oversegmentation. The grouping is based on the criteria of mean colour difference, edge strength of the shared borders and colour standard deviation of merged initial segments. The grouped segments are then evaluated using shape criteria in order to extract road parts. Results on some test images show that the approach provides reliable road parts. Concluding remarks are given at the end to point out further investigations concerning the evaluation of the road segments and their use in database assessment.
ASJC Scopus Sachgebiete
- Informatik (insg.)
- Maschinelles Sehen und Mustererkennung
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
2007. 51-56 Beitrag in Photogrammetric Image Analysis, PIA 2007, Munich, Deutschland.
Publikation: Konferenzbeitrag › Paper › Forschung › Peer-Review
}
TY - CONF
T1 - Road extraction in suburban areas based on normalized cuts
AU - Grote, A.
AU - Butenuth, M.
AU - Heipke, C.
PY - 2007
Y1 - 2007
N2 - This paper deals with road extraction of high resolution aerial images of suburban scenes based on segmentation using the Normalized Cuts algorithm. The aim of our project is the extraction of roads for the assessment of a road database, however, this paper is restricted to road extraction. The segmentation as our basic step is designed to yield a good division between road areas and the surroundings. We use the Normalized Cuts algorithm, which is a graph-based approach that divides the image on the basis of pixel similarities. The definition of these similarities can incorporate several features, which is necessary for the segmentation in complex surroundings such as built-up areas. The features used for segmentation comprise colour, hue, edges and road colour derived with prior information about the position of the centerline from the database. The initial segments have to be grouped due to an enforced oversegmentation. The grouping is based on the criteria of mean colour difference, edge strength of the shared borders and colour standard deviation of merged initial segments. The grouped segments are then evaluated using shape criteria in order to extract road parts. Results on some test images show that the approach provides reliable road parts. Concluding remarks are given at the end to point out further investigations concerning the evaluation of the road segments and their use in database assessment.
AB - This paper deals with road extraction of high resolution aerial images of suburban scenes based on segmentation using the Normalized Cuts algorithm. The aim of our project is the extraction of roads for the assessment of a road database, however, this paper is restricted to road extraction. The segmentation as our basic step is designed to yield a good division between road areas and the surroundings. We use the Normalized Cuts algorithm, which is a graph-based approach that divides the image on the basis of pixel similarities. The definition of these similarities can incorporate several features, which is necessary for the segmentation in complex surroundings such as built-up areas. The features used for segmentation comprise colour, hue, edges and road colour derived with prior information about the position of the centerline from the database. The initial segments have to be grouped due to an enforced oversegmentation. The grouping is based on the criteria of mean colour difference, edge strength of the shared borders and colour standard deviation of merged initial segments. The grouped segments are then evaluated using shape criteria in order to extract road parts. Results on some test images show that the approach provides reliable road parts. Concluding remarks are given at the end to point out further investigations concerning the evaluation of the road segments and their use in database assessment.
KW - Grouping
KW - Normalized cuts
KW - Road extraction
KW - Segmentation
KW - Urban areas
UR - http://www.scopus.com/inward/record.url?scp=84882990324&partnerID=8YFLogxK
M3 - Paper
AN - SCOPUS:84882990324
SP - 51
EP - 56
T2 - Photogrammetric Image Analysis, PIA 2007
Y2 - 19 September 2007 through 21 September 2007
ER -