Skip to Main Content
A vision-based lane-departure warning system which adapts the variation of brightness is developed in this paper. By sequentially applying the proposed processing techniques as lane strengthening, dichotomy, and smoothing filtering, an efficient decision rule is then applied to the processed image to realize the lane-departure warning system. Moreover, according to the variation of the brightness by reading the output of a photo-resistance sensor, the adaptive threshold for dichotomy is achieved. The present system effectively achieves real-time lane-departure warning with 10 decisions per second and 97.4% correct detection rate as driving angles varied larger than ±5° in the experiments.