基于改进欧式聚类的激光雷达目标检测

Lidar Target Detection Based on Improved Euclidean Clustering

针对传统欧式聚类算法对距离阈值敏感,易造成聚类目标过分割或欠分割的问题,提出了一种改进后的欧式聚类算法.改进后的算法针对合理的大距离阈值,在传统的欧式聚类搜索过程中,对聚类目标和干扰目标的激光点设定不同的激光点权值,去除搜索过程中干扰目标的激光点,较好地解决了大距离阈值聚类时目标的欠分割问题.同时,以大距离阈值避免了聚类目标的过分割现象.仿真结果表明,改进后的欧式聚类算法在一定范围的大距离阈值区间内都有较好的聚类效果,降低了传统欧式聚类算法距离阈值选取的难度.

To solve the problem that the traditional Euclidean clustering algorithm is sensitive to the distance threshold,which easily leads to over-segmentation or under-segmentation of the clustering target,an improved Euclidean clustering algorithm was proposed.The improved algorithm was aimed at the reasonable long-distance threshold.In the traditional Euclidean clustering search process,different laser point weights were set for the laser points of the clustered target and the interfering target,and then the laser points of the interfering target were removed in the search process,which better solves the problem of under-segmentation of the target in the long-distance threshold clustering.Meanwhile,the over-segmentation phenomenon of clustering objects was avoided by using a large distance threshold.Simulation results show that the improved Euclidean clustering algorithm has a good clustering effect in a certain range of large distance threshold interval,which reduces the difficulty of selecting distance threshold of traditional Euclidean clustering algorithm.