Lane Detection in Urban Environments Using Models of Probability and Evidence Theory

摘要

This contribution describes two approaches to video-based inner-city lane detection. The approaches address the particular challenges of inner-city traffic scenarios by using a probabilistic and evidence-theoretical formulation, respectively, in combination with a flexible lane model. Local orientation and anisotropy in the video image constitute the feature. The probability densities and the mass functions are formulated, fusion strategies discussed and results presented.