【摘要】 标准的多模型粒子滤波检测前跟踪技术是在低信噪比环境下检测与跟踪机动性的微弱目标的有效手段。但是由于其采用固定的运动模型数量,当运动模型数量过大时,模型之间的竞争会导致性能的下降。针对此问题,利用道路信息提出了一种变结构的多模型粒子滤波检测前跟踪算法。在每一时刻,根据目标的估计状态和挖掘的道路信息自适应地更新和改变运动模型集以能够选择更加有效的模型集,同时减少了模型数量,并且利用道路信息对目标的运动状态进行约束和限制。最后通过Monte Carlo仿真实验表明,基于文中所提出的算法在检测跟踪性能方面明显优于标准的多模型粒子滤波检测前跟踪算法。更多还原

【Abstract】 The standard multiple model(MM) particle filter(PF) track-before-detect(TBD) algorithm(MM-PFTBD) with a fixed structure and number of movement models has been become a popular and efficient approach to detect and track the maneuverable weak target in the much lower signal-to-noise ratio(SNR).But the performance of the detection and tracking is degraded owing to the competition among the movement models when the number of movement models is great.Then a novel TBD algorithm with variable structure MM-PF approach(VS-MM-PFTBD) is presented by utilizing the road information.The model set not only depending on the target state but also relating to the road information available is updated and varies adaptively so as to choose the effective model set and reduce the number of the movement models.In addition,target velocity constraints can be applied effectively along the directions of the different roads.Finally,the simulation results show that the VS-MM-PFTBD outperforms the standard MM-PFTBD using a fixed model structure without the road information.更多还原