SVM-based road verification with partly non-representative training data

Ziems, Marcel; Heipke, Christian; Rottensteiner, Franz

doi:10.1109/jurse.2011.5764713

Cited by 2 publications

(2 citation statements)

References 16 publications

(18 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thus, the value reflects the density of the training data; cf. (Ziems et al, 2011b). Intersection: This module is based on the method developed by Youn et al (2008) for road extraction in urban areas.…”

Section: Verification Modulesmentioning

confidence: 99%

Multiple-Model Based Verification of Road Data

Ziems

Breitkopf

Heipke

et al. 2012

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

ABSTRACT:This paper describes a semi-automatic system for road verification based on high resolution imagery and 3D surface models. Potential update regions are identified by an object-wise verification of all existing database records. The proposed system combines several road detection and road verification approaches from current literature to form a more general solution. Each road detection / verification approach is realized as an independent module representing a unique road model combined with a corresponding processing strategy. The object-wise verification result of each module is formulated as a binary decision between the classes "correct road" and "incorrect road". These individual decisions are combined by Dempster-Shafer fusion, which provides tools for dealing with uncertain and incomplete knowledge about the statistical properties of the data. For each road detection / verification module a confidence function for the result is introduced that reflects the degree of correspondence of an actual test situation with an optimal situation according to the underlying road model of that module. A comparison with results from an EuroSDR test on road extraction demonstrate the strengths and limitations of the method.

show abstract

Section: Verification Modulesmentioning

confidence: 99%

Multiple-Model Based Verification of Road Data

Ziems

Breitkopf

Heipke

et al. 2012

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

show abstract

“…Thus, the value reflects the density of the training data. The parameters a and b for the sigmoid function are selected so that the term becomes 0.9 for d(z,υ)=0 and 0.1 for d(z,υ)= d. A more comprehensive description is given in (Ziems et al, 2011).…”

Section: Edge Direction Analysismentioning

confidence: 99%

Multiple-Model Based Update of Belgian Reference Road Data

Ziems

Beyen²,

Mueller

et al. 2012

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

ABSTRACT:This paper describes a semi-automatic system for road update based on high resolution orthophotos and 3D surface models. Potential update regions are identified by an object-wise verification of all existing database records, followed by a scene-wide detection of redevelopment regions. The proposed system combines several road detection and road verification approaches from current literature to form a more general solution. Each road detection / verification approach is realized as an independent module representing a unique road model combined with a corresponding processing strategy. The object-wise verification result of each module is formulated as a binary decision between the classes "correct road" and "incorrect road". These individual decisions are combined by Dempster-Shafer fusion, which provides tools for dealing with uncertain and incomplete knowledge about the statistical properties of the data. For each road detection / verification module a confidence function for the result is introduced that reflects the degree of correspondence of an actual test situation with an optimal situation according to the underlying road model of that module. Experimental results achieved with data from the national Belgian road database in a test site of about 134 km² demonstrate the potential of the method.

show abstract

SVM-based road verification with partly non-representative training data

Cited by 2 publications

References 16 publications

Multiple-Model Based Verification of Road Data

Multiple-Model Based Verification of Road Data

Multiple-Model Based Update of Belgian Reference Road Data

Contact Info

Product

Resources

About