The LaneLok algorithms use a monochrome CCD video camera to detect lane markers on the road. These algorithms also estimate the lane boundary position the vehicle heading error the lateral offset and the look-ahead road curvature. The current LaneLok algorithms can identify lane markers and some obstacles up to only 40 feet. The lane control and collision warning functions of advanced automotive systems need a range of at least 100 feet for robust performance. A new algorithm was developed for extending this range to 100 feet using data from a single camera. Image data from the extended range was centered around the vertical vanishing point coordinate and enlarged in size using image processing techniques. A template matching algorithm was used to estimate lane boundaries of the near range (40 feet) and of the enlarged image. The two lane boundaries were then concatenated by a polynomial fit at the end points thus extending the lookahead range to 100 feet. For realtime implementation new edge detectors were developed for saving processing time and electronic hardware.
Surender K. Kenue,
"LANELOK: an algorithm for extending the lane sensing operating range to 100-feet", Proc. SPIE 1388, Mobile Robots V, (1 March 1991); doi: 10.1117/12.48089; https://doi.org/10.1117/12.48089