On vision-based lane departure detection approach

Abstract

This paper analyzes the process of lane departure detection approach. In our vision-based lane departure detection system, we use a single camera as input. In this paper, we discuss how to detect the lanes marking on the road and get the relationship between vehicle and road. Some measurements are derived to calculate TLC (time to lane crossing), for measuring the position of the vehicle relative to the lanes. Besides, the forward-looking predicting mode is introduced to establish the relationship among driver, vehicle, and road. Further more, the criterion can be got to tell whether there is an unintended lane departure. Simulations show that our vision-based lane departure approach does provide an effective alarm when the state of driver goes wrong.