Road Boundary Detection in Challenging Scenarios

摘要: This paper presents a new approach for automatic road detection in traffic cameras. The technique proposed here detects the dominant boundary and estimates vanishing point images captured by cameras under wide range of lighting environmental conditions, e.g., unlit highways at night, etc. starts segmenting scene into number superpixel regions. contours these regions are used to generate large edges which organized clusters co-linearly similar sets using hierarchical bottom up clustering. A confidence level is assigned each cluster statistical best chosen. Pairs with high levels then ranked filtered according image perspective activity. top pair selected as boundary. tested on real world dataset collected from Ontario 401 surveillance system. Experimental results demonstrate distinct speedup improvement accuracy detecting challenging scenarios compared state art Gabor filter based technique.