摘要
提出了一种高速公路车道线检测与跟踪算法:初始检测时,首先估计消失点位置;再将图像特征点向消失点投影,通过检测投影直方图波峰来识别车道线;然后利用车道宽度和车道线强度特征滤除与车道线平行的伪车道线;跟踪时,采用卡尔曼滤波缩小搜索区域。试验结果表明:该算法抗干扰能力强,跟踪算法运行速度大约为50 Hz。
A real time approach that detects and tracks lane markers on highway is presented.Firstly,the location of vanishing point is estimated in the initial detection step.Then,the feature points on the image are projected to the vanishing point.The lane markers are detected according to the peaks in projection histogram.Finally,the fake lane markers that are parallel to the real lane markers are filtered out by means of the lane width feature and the lane markers strength feature.In the tracking step,the Kalman filter is proposed to narrow the search region.The result of experiments indicates that this approach has a strong ability to resist various disturbing factors and the tracking algorithm runs at a rate of 50Hz.
出处
《装甲兵工程学院学报》
2013年第6期61-64,84,共5页
Journal of Academy of Armored Force Engineering
基金
军队科研计划项目
关键词
车道线检测
消失点
卡尔曼滤波
lane markers detection
vanishing point
Kalman filter