摘要
为了让智能汽车辨识可行驶区域,道边检测是前提。使用多线激光雷达,通过对大量道边点数据进行分析,提出一种道边检测与跟踪算法。首先,通过分析扫描点特征,建立多阈值筛选算法,提取出有效道边点集;其次,采用基于K-means改进的聚类算法对有效道边点集进行聚类分析,得到左、右两侧的道边点集;最后,使用最小二乘法拟合得到左右道边。经过实际验证,该算法道边检测准确,处理每帧数据平均仅需34ms。
As the precondition of autonomous driving,road edge detection was crucial for intelligent vehicles to recognize the free driving space.By analyzing plenty of road edge point data,we propose a new road edge detection and tracking algorithm.Firstly,a multi-threshold algorithm was used to extract valid road edge point by analyzing the characteristic of scanning point.Secondly,an improved algorithm based on K-means algorithm was used to classify road edge into right and left road edge points.Finally,two road edges were fitted by least square algorithm.Experimental results showed that the algorithm can accurately detect road edge,which only costs 34 ms to process one data frame.
作者
高栋南
GAO Dong-nan(School of Optical-Electrical and Computer Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China)
出处
《软件导刊》
2018年第7期23-26,共4页
Software Guide
基金
国家自然科学基金项目(61374197)