Novel Multiple Lane Detection Based on Dense Vanishing Point Estimation

The detection of multiple curved lanes is still a challenge for Driver Assistance Systems (DAS) today, due to interferences such as road markings and shadows casted by road side structures and vehicles. The vanishing point (Vp) contains the global information of the road image. Hence, Vp based lane detection algorithms are quite insensitive to interference. However, when curved lanes are assumed, Vp shifts with respect to the rows of the image. In this paper, a Vp for each individual row of the image is estimated by first extracting a Vpy (vertical position of the Vp) for each individual row of the image from vdisparity. Then, based on estimated Vpys a 2D Vpx accumulator is efficiently formed. Thus, by globally optimizing this 2D Vpx accumulator, globally optimum Vps for road image is extracted. Then, estimated Vps are utilized for multiple curved lane detection on non-flat surfaces. The resultant system achieves a detection rate of 99% in 1277 frames of 5 stereo vision test sequence.