中文摘要：

针对机动目标跟踪系统建模中的非线性问题，提出一种基于容积卡尔曼滤波（CKF）的雷达与红外传感器融合算法。考虑到被估计系统对目标跟踪算法实时性与精度的要求，在容积滤波框架下构建了集中式量测融合（CMF）和分布式状态融合（DSF）两种结构形式。CMF结构采用最优加权方法，首先对雷达和红外两种异类传感器的方位角度量测信息进行融合，并将其与融合后的雷达径向距量测构建新的量测数据，进而通过CKF算法对机动目标进行跟踪。DSF结构则首先对雷达量测中径向距信息进行加权融合，并将融合结果作为红外传感器的虚拟径向距量测，以实现红外量测的扩维处理，进而对每组量测数据应用CKF进行分布式并行加权融合，获得目标运动状态的最终估计。仿真场景中，对两种融合方法的性能进行比较，理论分析与仿真实验验证了算法的可行性与有效性。

英文摘要：

Due to the nonlinear problem existing in the system modeling of maneuvering target tracking, a novel heterogeneous sensor fusion algorithm based on cubature Kalman filter（CKF） is proposed. Considering the accuracy and real time in target tracking realization, two realization structures, centralized measurements fusion and distributed state fusion, are respectively constructed under the framework of CKF. In the architecture of centralized measurements fusion,the optimal weighted method is used to deal with the azimuth observations from radar and infrared sensors, and then the combination of the fusion result and the radial distance information fused from radar is taken as a group of new measurements. Fi- nally,the CKF algorithm is used for maneuvering target tracking based on new measurements. In the framework of distributed state fusion,in order to achieve the dimension augmenting of the infrared sensor measurements, the radial distance information from radar is fused, and the fusion result is defined as the virtual distance measurement of all infrared sensors, and then the CKFs based on different groups of measurements are applied to obtain the final state estimation of the maneuvering target by means of the distributed weighted fusion. The performance of the two fusion architectures is compared in the simulation scene, and the theoretical analysis and experimental results show the feasibility and efficiency of the proposed algorithm.