Real-time and robust method for unstructured roads detection
一种实时鲁棒的非结构化道路检测算法*

This paper proposed an unstructured road detection algorithm based on block-segment Gaussian mixture model and B-spline curve model, called BSGMM_BS. First modeled the road and off-road regions by Gaussian mixture color model. Then introduced a method called block-segment method to extract the road edge points. Finally fitted two boundaries of the lane by the B-spline curve in which the best control points were searched by the least square method. Field tests and comparison experiments show that the algorithm is real-time and is robust against shadows, illumination variations.