摘要
建立了直线道路模型,利用图像处理技术对道路图像进行了预处理,采用抽行扫描的方法提取出道路标识线的特征点,并对这些特征点采用线性回归的方法得到道路标识线的方程,为提高标识线识别的鲁棒性,在对特征点的优劣性进行判别的基础上,采用高准确度的特征点对道路图像中的车道标识线进行识别,进而选定车辆与道路的方向偏差、位置偏差和方向偏差的变化率3个参数来判断车辆在道路上的位置,以此作为智能车辆导航的依据,大量的实验表明文该方法是有效、可行的。
Based on linear road model, it introduces the method to process road image, that uses the scanning extracted lines of road image to find lane mark character points and lane mark line＇ s equation based on linearity regression, evaluates the quality of character points, and applies the character points with high quality to identify the lane mark lines. This method can select the direction wrap, the position wrap between vehicle and road, and diversification rate of direction wrap to decide the position of the vehicle on the road. The experiment shows that the method is sufficient.
基金
山东省自然科学基金资助项目(Q99F11)
关键词
图像处理
道路标识线
抽行扫描
导航参数
Image Process
Lane Marking Line
Scanning Extracted Lines, Navigation Parameter
