数据融合系统中并行目标识别的研究与实现

分析了将情报侦察数据融合系统中目标识别中心并行化的可行性，提出了基于航迹的任务划分策略，采取集中式动态负载平衡技术设计并实现了基于消息传递接口（MPI）的并行目标识别中心。同时给出了Dempster-Shafer证据理论在适当条件下的一种简化形式及其在并行环境下的应用方法。最后给出融合系统并行目标识别中心的性能测试结果，在保证识别可信度的基础上大大提高了处理速度，解决了融合系统的性能瓶颈问题。

Research and Implementation of Parallel Object Recognition in Data Fusion System

This paper deals with the parallel object recognition of the information reconnaissance and data fusion systems, analyses its feasibility provides a strategy of partitioning based on tracks, designs and implements the parallel object recognition center based on the message passing interface with a dynamic load balancing technology. While a predigested form of the dempster-shafer theory of evidence under some special situations is presented and how to use it is told. Finally, it gives the results of the performance test to the parallel object recognition center. It can be seen that the speed of recognition is improved while ensuring the reliability of the recognition, which solves the key performance problem of the fusion system.