摘要
为进一步提高安全驾驶结构化道路车道线检测的实时性和鲁棒性,该文提出了一种基于TMS320DM6446嵌入式处理平台的车道线快速检测算法。算法在所设置的感兴趣区域,融合最优阈值化及区域生长法获取车道线候选点集,采用连通聚类法获取候选点集聚类中心以提取车道线候选点,紧接着对候选点做进一步筛选和分类从而获得有效特征点,最后对有效特征点进行最小二乘拟合以实现车道线检测。实验结果表明,该算法能够快速检测车道线,为车辆偏离警告系统的有效决策提供了有利的前提条件。
To further enhance safe driving structured way of robust and real-time lane marking detection is proposed based on TMS320DM6446 embedded processing platform for rapid detection algorithm. Algorithm in the set region of interest, the integration of the optimal threshold and region growing method to obtain the lane candidate point sets, the use of connected clustering method for clustering center of the candidate point sets to extract the lane mark candidate points, followed by the candidate points further screening and classification in order to obtain the effective feature points, finally detect the lane by the least squares fitting with the effective feature points. The experimental results show that the algorithm can quickly detect the lane, provided a favorable precondition for effective decision-making for the Vehicle departure warning system.
出处
《杭州电子科技大学学报（自然科学版）》
2013年第2期37-40,共4页
Journal of Hangzhou Dianzi University：Natural Sciences
基金
浙江省自然科学基金资助项目(Y1100086
Y1100803)
浙江省创新团队资助项目(R50009)
关键词
车道线
候选点
有效特征点
最小二乘法
lane marking
the candidate points
effective feature points
least square
