使用高斯混合滤波器的机动目标跟踪

摘要

实际目标跟踪过程中，被跟踪目标的状态与类型都是不确定的。使用运动学传感器与属性传感器分别获取的目标状态量测信息与特征量测信息，给出了目标状态与类型不确定性的联合状态类型概率密度函数表示，并推导了线性高斯假设下的系统模型为高斯混合模型。根据这一性质，引入高斯混合滤波器，实现了机动目标的有效跟踪。在仿真分析中，通过对比3种算法的跟踪结果，进一步验证了使用高斯混合滤波器在机动目标跟踪过程中的有效性。%In actual target tracking process,the state and the type of target being tracked are uncertain.By using the state information and the feature information obtained respectively by kinematic sensor and attribute sensor,we obtained a joint state/type probabilistic density function for describing the uncertainty of target state and target type,and proved that the system model under the linear Gaussian assumptions is a Gaussian mixture model.Based on this property,a Gaussian mixture filter was introduced to implement maneuvering target tracking effectively.In simulation and analysis,the effectiveness of the tracking algorithm using Gaussian mixture filter was validated by comparing it with two other classic tracking algorithms.