联合Metropolis算法和MDL准则的密集多目标分辨算法  

Multiple unresolved targets distinguish algorithm by combination of the Metropolis algorithm with MDL criteria

作　　者：李振兴[1] 刘进忙[1] 周政[1] 郭相科[1] 李延磊[1] 

机构地区：[1]空军工程大学防空反导学院,陕西西安710051

出　　处：《西安电子科技大学学报》2014年第3期174-180,共7页Journal of Xidian University

基　　金：国家自然科学青年基金资助项目(61102109);陕西省自然科学基金资助项目(2010JM8013);空军工程大学防空反导学院研究生科技创新基金资助项目(HX1112)

摘　　要：针对单脉冲雷达处理同一分辨单元内密集多目标回波时收敛较慢的问题,提出了一种将Metropolis算法和最小描述长度(MDL)准则相结合的密集多目标分辨算法.构建Metropolis算法中的更新函数和迭代规则,促使待估参数的不同马尔可夫链间的融合;根据融合判定规则选取抽样样本,估计出对应的目标参数;利用MDL准则实现密集目标的准确分辨.该算法不仅提升了收敛速度,而且具有较高的参数估计精度,提高了算法在多目标下的分辨性能.仿真结果验证了所提算法的有效性和可行性.When multiple targets＆#39; echoes fall into the same resolution cell of monopulse radar , the echoes are distinguished at a slow rate of convergence using the available processing algorithm . Aiming at this problem , a new processing algorithm is presented , which combines the Metropolis algorithm with minimum description length （ MDL） criteria . Constructing the update function and the rule of iteration can improve the convergence of the different markov chains of parameters estimation . Then , by using the fusion decision rule , valid samples are selected to calculate the parameters . Lastly , the numbers of multiple unresolved targets are calculated by using minimum description length criteria . The proposed algorithm has not only a higher rate of convergence , but also a better resolution capability under multiple targets with a more accurate parametric estimation . The efficacy and feasibility of the method are verified by simulation .

关 键 词：密集多目标 分辨算法 Metropolis算法 MDL 准则 马尔可夫链 

分 类 号：TN95[电子电信—信号与信息处理]