A Robust Approach of Lane Detection Based on Machine Vision

The lane detection is a key component of the intelligent transportation systems (ITS). We present a robust approach of lane detection based on machine vision. First, we present the lane model and region of interest (ROI) of the road image. Then, we propose the edge detection approach of the road image based on gray value grade. After that, we illustrate how to remove the interference points in the previous processed image; meanwhile, we describe how to gather the valid points. At last, we employ the coarse Hough transform to estimate the parameter values of the lanes. We present how to use Kalman filter to refine the estimation results. The field tests are carried on a local high-way and the experimental results show that the suggested approach is very reliable.