Vision-based Lane Detection Algorithm

Intelligent Vehicle (IV) is a new research field in many counties, and, vision-based structured road is a important research direction. Structured road detection is mainly depending on the detection of the lanes on the road and to confirm the boundary of the road. The final goal is the autonomous calm, safe drive for intelligent vehicle in any time and without people's help.In this paper, we mainly direct to the characters of structured road, we did some work on how to extract lane information in road images efficiently. We did our work on three sides:Based on the characters of road images, we researched on how to preprocess road images.For straight lanes, we give out a Hough Transform based algorithm, the algorithm can also estimate the lane is a real lane or a dashed lane.For curve lanes, we give out a slope closing algorithm.Road images are complicated, this is because the complication of the road environments. So, we did some hypothesis on the road, we combine these hypothesis and our detection algorithm.From the experiment, we can see that our algorithms are valid, but the algorithms have many drawbacks also.