论文部分内容阅读
数据融合是一个对来自多源的数据和信息进行互联、相关、组合处理以求得精确位置和识别估计的过程。本文考虑其中的配准问题,它是数据融合系统求得精确估计和修正系统误差所必需的预先处理。文中介绍了配准的精确极大似然算法(EML)。该算法通过两步递归最优化方法来实现,并采用改进的高斯—牛顿法来确保算法的快速收敛性。文中研究了该算法的统计性能,其中包括对一致性和有效性的讨论。我们特别地推导并给出了渐近协方差和克拉默—劳边界(CRB)的显式。最后,用仿真和实际的多雷达数据来评估该算法的性能。
Data fusion is a process of interconnecting, correlating and combining data and information from multiple sources to find the exact location and identify the estimation. This paper considers the registration problem, which is necessary for the data fusion system to obtain the accurate estimation and correction of systematic errors. The paper introduces the exact maximum likelihood algorithm (EML) for registration. The algorithm is implemented by two-step recursive optimization method, and the improved Gauss-Newton method is adopted to ensure the fast convergence of the algorithm. In this paper, the statistical performance of the algorithm is studied, including the discussion of consistency and validity. In particular, we derive and give the explicit asymptotic covariances and the Kramer-Law boundary (CRB). Finally, the performance of this algorithm is evaluated using simulated and actual multi-radar data.