Research On Target Detection And Tracking Algorithm Based On Four-Layer Lidar

With the rapid economic and social development,traffic congestion,frequent traffic accidents,and environmental pollution have become more serious.Intelligent vehicle is one of the important ways to solve these problems.Intelligent vehicle systems mainly include environmental awareness,planning decisions,and control execution.In environment perception,four-layer lidar is widely used because of its advantages such as moderate data volume,high measurement accuracy,and low environmental impact.This paper studies the target detection and tracking algorithm based on the four-layer lidar,and mainly carries out the following work:First,the installation,calibration and coordinate system conversion methods of the four-layer lidar are introduced.The filter method based on the difference between the maximum and minimum heights of the grid map was chosen to filter the original data of the four-layer lidar,and the filtering of noise and ground points in data is realized.Secondly,according to the distribution characteristics of the four-layer lidar data and comparing with different clustering algorithm,DBSCAN(Density-Based Spatial Clustering of Applications with Noise)algorithm was selected as the target detection algorithm.In view of the fixed input parameters,the parameters of the DBSCAN clustering algorithm are adaptively determined by the nearest neighbor scale deductive method.The minimum enveloping rectangle extraction algorithm based on ergodic method is used to extract the features of each target with a rectangular frame.Thirdly,An improved nearest neighbor algorithm based on multi-feature matching is proposed to complete the fast matching of two adjacent lidar data.A multi-feature matching function is constructed to replace Euclidean distance.The tracking of dynamic targets is realized by using Kalman filter and tracking management.Finally,liada rvisua linterface developed under Ubuntu 14.04,and Hunan University's LIFAN-?,as the experimental platform,is used to do massive experiments.These experiments are used to test the effectiveness of the proposed filtering,clustering algorithm and tracking algorithm.Experimental results show that the proposed target detection and tracking algorithm can extract target feature information stably and reliably and track dynamic targets.