一种有效融合高冲突证据的目标身份识别方法

摘要

The Dempster-Shafer theory(DST) has been widely used for the target identity identification. However, the failure in dealing with conflicting evidence by Dempster combination rule has been the focus of attention all the time. In order to fuse the highly conflicting evidence effectively, a target identity identification method based on the combination of the Dezert-Smarandache theory (DSmT) and the DST is proposed. The mothed adopts the fusion mode of single frame firstly and multiframe later. In the single frame fusion the DSmT is used to solve the high conflicts, while in the multiframe fusion the DST is chosen for ensuring the good performance of fast convergence speed and light computation load. The validity and the applicability of the method is demonstrated by the simulation results finally.%DS理论(Dempster-Shafer theory,DST)已广泛应用于目标身份识别中,但是Dempster组合规则在处理冲突证据时的失效问题一直是研究的热点.为能在目标身份识别中有效融合高冲突的证据,提出DSm理论(Dezert-Smarandache theory,DSmT)和DST的混合方法.该方法采用先单帧后多帧的融合模式,在单帧融合中采用DSmT,化解可能的高冲突证据;在多帧融合时采用DST,保证了快速收敛性和较小的计算量.仿真分析验证了该方法的有效性和适用性.