融合位置和姿态信息的模型概率估计

Model Probability Estimation Based on Information Fusion of Position and Attitude

为了利用目标的姿态信息来提高机动目标跟踪的精度,提出一种融合位置和姿态信息的模型概率估计算法。该算法采用模糊关联的方法,实现了基于姿态信息的模型概率估计;根据离散点过程滤波的基本理论,对其进行滤波;采用贝叶斯推理方法,对来自姿态的模型概率和来自位置的模型概率进行了信息融合,并将融合后的模型概率应用到IMM算法。仿真表明,改进的IMM在保持传统IMM实时性的基础上提高了跟踪精度,从而验证了姿态信息对目标跟踪的辅助作用。

In order to use attitude information to improve the precision of maneuvering target tracking, an estimating algorithm of model probability was proposed based on information fusion of position and attitude. Firstly, the algorithm adopted fuzzy association method to estimate model probability based on attitude information. Then, the estimated model probability was filtered according to the theory of discrete point process filtering. At last, model probability from attitude and that from position were fused based on Bayes inference method, and the fused model probability was introduced into IMM algorithm. Simulations show that the improved IMM has as good real-time performance as traditional IMM but better tracking precision, which proves the assistance of attitude to target tracking.