Graphics Reference
In-Depth Information
Table 10.1
Lane analysis: Illustrative research studies
Lane Analysis Framework
Basic formulation
Algorithms
McCall [
10
]
Directionality coupled with intensity transitions of lanes
Steerable filters for lane feature extraction
2006
Hough transform, IPM &Road model for outlier removal
Lane positions can be predicted over time based on vehicle
movement
Steerable filter response to differentiate lane markings and
circular reflectors
Lane markings show directionality, circular markers do not
Kalman filtering for lane tracking
Ego-vehicle position estimation using lane positions
Cheng [
3
]
Extension of formulation as in [
10
] to omnidirectional cameras
Extension of lane estimation methods in [
10
]
2007
Ego-vehicle position estimation using lane positions
Borkar [
1
]
Dashed lanes when averaged can give continuous lanes
Temporal blurring for lane feature extraction
2012
Lanes are brighter than lanes, especially at night
Adaptive thresholding
Lanes have two different gradients
Gaussian kernel template matching and RANSAC for outlier
removal
Lane positions can be predicted over time based on vehicle
movement
Kalman filtering for lane tracking
Gopalan [
5
]
Lanes have predetermined properties which can be learnt
Learning based method for lane feature extraction
2012
Lane positions can be predicted over time based on vehicle
movement
Particle filtering for lane tracking
Nedevschi [
12
] 2013
Lanes are brighter than road, have constant width
Gray level based feature extraction
Movement of lane markings is periodic across time
Periodic histograms for lane boundary classification
Double lanes detection
Ego-vehicle global localization using visual and GPS data
(continued)
