摘要
Lane departure warning system based on machine vision is a human decision-make like solution to avoid lane departure fatalities with low cost and high reliability. In this paper, the model of vision-based lane departure warning system and the realization is described at first. Then the method of lane detection is illustrated, which is composed of three steps: image preprocessing, binary processing and dynamical threshold choosing, and linear-parabolic model fitting. After that, the solution of how to perform the departure decision-making is proposed and demonstrated. Unlike the usual TLC([1]) and CCP([2]) methods, the angles between lanes and the horizontal axis in captured image coordinate are used as the criterion for lane departure decision-making. At last the experiments are implemented by use of all the steps; the results indicate the efficiency and feasibility of the solution.
- 出版日期2008
- 单位东南大学