基于道路特征点建立前方可行区域的一种方法

Building safe region based on road characters

本文提出了一种在前方车道图像中建立高分辨率的ROI(Region Of Interest)即感兴趣区域,并在其中建立搜索窗口以提高车道识别精度的方法,实验证明这种方法能够更好的处理非理想路况的各种不确定因素,除了能更精确的追踪车道标志线之外,还能更好的应用于前方图像出现噪声、阴影以及路面反光等不利情况。由于高分辨率ROI的建立,在不增加运算时间的前提下,识别精度却更加提高,同时整个运算过程的计算量大大减少。

The paper brings forward a method which sets highly differentiate region of interest（ROI） on the frontage roadway picture,and sets search windows to improve the precision of roadway recognition,the experiment prove that this method can dispose every kind of uncertainty factors very well on road surface if the road surface condition is not favorably,and set real time dynamic search windows in ROI.Besides,it can track down the lane more accurately.This method also can be applied in frontage image,including shadow condition and crooked road condition very well.As the setting of ROI,the computing time wouldn＇t be increased,the differentiate of recognition be increased more,and the compute of search process can be reduced greatly.