摘要
针对雷达信号处理中的点迹凝聚算法在应对多目标、群目标检测时的实时性能不佳,对点迹凝聚算法的工程实现提出了优化措施。对于常规实现中的耗时操作采用预排序和哈希映射的方式优化,降低了算法实现的时间复杂度。为使算法适配多核处理器,提出了2种多线程的实现方案。对方案进行性能对比,分析了不同优化方案适用的场合。仿真结果证明在目标数目达到一定规模后,提出的方案相对于常规实现有着近百倍的加速,有效解决了点迹凝聚算法处理大量点迹时实时性不佳的问题。
In order to solve the problem of poor real-time performance caused by a large number of data when dealing with multi-target and group target detection,this paper proposed an optimization measures for the engineering implementation of the plots centroid algorithm.For the time-consuming operation in the conventional implementation,the data structure of pre-sorting and HashMap was used to optimize,which reduces the time complexity of the algorithm implementation.Meanwhile,in order to adapt the algorithm to multi-core processor,two multi thread implementation schemes were proposed.Finally,the performances of the schemes were compared,and the application occasions of different optimization schemes were analyzed.The results demonstrate that when the number of points reaches a certain scale,the optimal solution has nearly 100 times faster than the conventional implementation,which solves the problem of poor real-time performance when processing a large number of plots.
作者
吴春林
曹运合
王蒙
WU Chunlin;CAO Yunhe;WANG Meng(Qingdao Branch of Naval Aviation University, Qingdao 266041, China;Xidian University, Xi’an 710071, China;Beijing Electro-Mechanical Engineering Institute, Beijing 100074, China)
出处
《兵器装备工程学报》
CSCD
北大核心
2021年第9期196-201,共6页
Journal of Ordnance Equipment Engineering
基金
国家自然科学基金项目(61771367)。
关键词
点迹凝聚
快速排序
哈希映射
多线程优化
plots centroid algorithm
quick sort
HashMap
multi thread implementation scheme