基于模型混合的多分辨率多模型跟踪算法

在原有的多分辨率多模型算法的基础上引入了滤波器混合的方法,提出了一种新的机动目标跟踪算法.原多分辨率算法在多分辨率框架下应用硬切换多模型算法,改善了多模型算法在非机动段的跟踪性能,但是保留了硬切换算法的固有缺点.交互式多模型算法由于采用模型混合,避免了硬切换算法存在的问题,但是非机动段的跟踪性能不如多分辨率多模型算法.所提出的新算法综合了两者的优点,不仅解决了多分辨率多模型算法中存在的问题,而且其综合滤波性能优于交互式多模型算法.

Multi-Resolution Multi-Model Tracking Algorithm Based on Model-Mixing

A model-mixing approach has been introduced into the original multi-resolution multi-model algorithm and a new maneuvering target tracking algorithm based on multi-resolution multi-model method is proposed. The original algorithm adopts hard switching multi-model approach in the multi-resolution frame. It can improve the performance in the nonmaneuvering segment, but it remains the inherent defaults of the hard switching method. The interactive multi-model algorithm can avoid this default because of model-mixing, but its performance isn-t good during non-maneuvering. The new algorithm combines the advantages of the original multi-resolution algorithm and that of the interactive multi-model algorithm. It can not only solve the problems existed in the original algorithm, but also achieve to a better performance than IMM algorithm during the entire tracking procedure.