Robust curved road boundary identification using hierarchical clustering.

We develop a new method for automatic curved road boundary detection in images captured by traffic cameras. The proposed method combines data driven (edge segments) and model based (2nd degree polynomial models for road boundaries) techniques to identify dominant road boundary in images exhibiting extreme weather conditions, low visibility and poor lighting. The proposed method constructs a ranked list of possible road boundaries through agglomerative hierarchical clustering of edge segments. Each node in the hierarchical clustering is a potential road boundary. Top ranked road boundaries are paired with each other to identify potential road regions. The road regions are then ranked using appearance and perspective cues and the top ranked road region is used to construct the dominant road boundary in the image. We evaluate our method on a realistic dataset captured by traffic cameras managed by Ontario’s Ministry of Transportation.

Name thesis:degree_name

Master of Science (MSc)

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Ontario Institute of Technology

Year dc:date.issued

2013

Author dc:creator

• Parvin, Nurjahan

Advisors dc:contributor.advisor

• Pu, Ken
• Qureshi, Faisal

• Road boundary
• Hierarchical clustering

Language dc:language.iso

en

Handle dc:identifier.uri

https://hdl.handle.net/10155/363

OAI identifier oai:identifier

oai:ontariotechu.scholaris.ca:10155/363