基于可逆跳跃马尔可夫链蒙特卡罗方法的空间邻近目标红外像平面分辨

Resolution of Closely Spaced Objects via Infrared Focal Plane Using Reversible Jump Markov Chain Monte-Carlo Method

空间邻近目标在红外像平面的成像相互交叠,形成簇状像斑,为实现对各目标的有效跟踪与识别,必须对其进行分辨。提出了基于可逆跳跃马尔可夫链蒙特卡罗(RJMCMC)的空间邻近目标红外像平面分辨方法。在对多点源目标的红外像平面成像建模基础上,建立了基于贝叶斯推理的空间邻近目标红外像平面分辨框架,以可逆跳跃马尔可夫链蒙特卡罗方法实现了待估参数后验分布的计算,联合检测和估计出目标个数、各目标像平面投影位置和辐射强度参数。以中段弹道空间邻近目标的天基红外监视为例进行了仿真分析,结果表明,该方法能有效地分辨空间邻近目标中的目标个数和各目标投影位置和辐射强度。

The closely spaced objects（CSO） create blur pixel cluster on the infrared focal plane;therefore,in order to effectively track and identify each object of CSO,it is necessary for the sensor signal processing to resolve them.A novel method of CSO infrared resolution based on reversible jump Markov chain Monte-Carlo（RJMCMC） method is presented.The method firstly creates an infrared focal plane image model,then constructs a framework of Bayesian inference for CSO resolution,and subsequently uses the RJMCMC to perform computation of the parameters′ posterior distribution,ultimately the joint estimation of objects number is obtained,positions and intensities.Simulation with infrared sensor viewing midcourse CSO are carried out to test the performance of the method,and the results confirm the effectiveness of the method.